Azure AI Foundry (from BUILD 2025) with Yina Arenas

Maybe you've been involved in a Maybe you've been involved in a Microsoft 365 migration, optimizing your Microsoft 365 migration, optimizing your Microsoft 365 migration, optimizing your environment, maybe prepping for environment, maybe prepping for environment, maybe prepping for co-pilot, and then you've probably heard co-pilot, and then you've probably heard co-pilot, and then you've probably heard the name Sharegate. It rarely gets the the name Sharegate. It rarely gets the the name Sharegate. It rarely gets the spotlight, but everything runs on what spotlight, but everything runs on what spotlight, but everything runs on what they build. Mergers, restructures, they build. Mergers, restructures, they build. Mergers, restructures, tenant sprawl, it figures it out as they tenant sprawl, it figures it out as they tenant sprawl, it figures it out as they handle everything from migrating data handle everything from migrating data handle everything from migrating data from tenant to tenant to making sure from tenant to tenant to making sure from tenant to tenant to making sure that AI and co-pilot is ready to be that AI and co-pilot is ready to be that AI and co-pilot is ready to be rolled out securely and responsibly. rolled out securely and responsibly. rolled out securely and responsibly. They make chaos feel like business as They make chaos feel like business as They make chaos feel like business as usual. And Sharegate, it's the sidekick usual. And Sharegate, it's the sidekick usual. And Sharegate, it's the sidekick that helps them do it. Sharegate is an that helps them do it. Sharegate is an that helps them do it. Sharegate is an out-of-the-box tool that simplifies the out-of-the-box tool that simplifies the out-of-the-box tool that simplifies the chaos that it faces. Assess, migrate, chaos that it faces. Assess, migrate, chaos that it faces. Assess, migrate, optimize. It's that simple. Now, whether optimize. It's that simple. Now, whether optimize. It's that simple. Now, whether it's tenant to tenant, Microsoft it's tenant to tenant, Microsoft it's tenant to tenant, Microsoft Exchange, Google to M365, or getting Exchange, Google to M365, or getting Exchange, Google to M365, or getting visibility on your data hygiene in order visibility on your data hygiene in order visibility on your data hygiene in order to prep for co-pilot rollouts, ShareGate to prep for co-pilot rollouts, ShareGate to prep for co-pilot rollouts, ShareGate helps it migrate quickly, cleanly, and helps it migrate quickly, cleanly, and helps it migrate quickly, cleanly, and with minimal drama. Go to sharegate.com. with minimal drama. Go to sharegate.com. with minimal drama. Go to sharegate.com. It's one tool, fast migrations and It's one tool, fast migrations and It's one tool, fast migrations and co-pilot co-pilot co-pilot [Music] [Music] [Music] ready. Hey friends, this is a slightly ready. Hey friends, this is a slightly ready. Hey friends, this is a slightly different episode of Hansel Minutes, so different episode of Hansel Minutes, so different episode of Hansel Minutes, so I just want to give you a heads up. What I just want to give you a heads up. What I just want to give you a heads up. What I did was I recorded a talk that I gave I did was I recorded a talk that I gave I did was I recorded a talk that I gave at Microsoft Build 2025 with Yina Arenas at Microsoft Build 2025 with Yina Arenas at Microsoft Build 2025 with Yina Arenas where we talked about my podcast and in where we talked about my podcast and in where we talked about my podcast and in the middle of the talk I decided to make the middle of the talk I decided to make the middle of the talk I decided to make that talk the podcast. It might not be

that talk the podcast. It might not be that talk the podcast. It might not be super effective for folks that are audio super effective for folks that are audio super effective for folks that are audio only, but I'm going to make sure to only, but I'm going to make sure to only, but I'm going to make sure to include a link to the video in the show include a link to the video in the show include a link to the video in the show notes. But this is that recording notes. But this is that recording notes. But this is that recording delivered to you as a podcast where we delivered to you as a podcast where we delivered to you as a podcast where we talk about how to produce this podcast. talk about how to produce this podcast. talk about how to produce this podcast. Enjoy. Enjoy. Enjoy. Hi friends. Hi friends. Hi friends. Hello. Thank you for joining us today. A lot of Thank you for joining us today. A lot of people have worked very very hard to people have worked very very hard to people have worked very very hard to make uh some really cool stuff here. We make uh some really cool stuff here. We make uh some really cool stuff here. We have a page and a half of demos, demos, have a page and a half of demos, demos, have a page and a half of demos, demos, demos and just an hour to do it and it's demos and just an hour to do it and it's demos and just an hour to do it and it's all live. Yes. All right, let's do some all live. Yes. All right, let's do some all live. Yes. All right, let's do some uh let's do some stuff. Let's get uh let's do some stuff. Let's get uh let's do some stuff. Let's get rolling. Well, welcome, welcome, rolling. Well, welcome, welcome, rolling. Well, welcome, welcome, welcome. We're going to talk about guess welcome. We're going to talk about guess welcome. We're going to talk about guess what? what? what? AI. AI. AI. [Music] [Music] [Music] Actually, I Let me I just realized Actually, I Let me I just realized Actually, I Let me I just realized something. You know, I've never actually something. You know, I've never actually something. You know, I've never actually had you on my podcast. I know. Can I had you on my podcast. I know. Can I had you on my podcast. I know. Can I record this? We We've met like 10 years record this? We We've met like 10 years record this? We We've met like 10 years ago and we've talked about this so many ago and we've talked about this so many ago and we've talked about this so many times. Hang on. Let me um let me just times. Hang on. Let me um let me just times. Hang on. Let me um let me just hit record.

hit record. hit record. All right. All right. All right. Hi, I'm Scott Hansselman. This is Hi, I'm Scott Hansselman. This is Hi, I'm Scott Hansselman. This is another episode of Hansel Minutes. Today another episode of Hansel Minutes. Today another episode of Hansel Minutes. Today on the show we have Yina Arenus talking on the show we have Yina Arenus talking on the show we have Yina Arenus talking about Azure AI Foundry. Yes. All right. about Azure AI Foundry. Yes. All right. about Azure AI Foundry. Yes. All right. Thanks. Sorry. I'm going to keep that Thanks. Sorry. I'm going to keep that Thanks. Sorry. I'm going to keep that running. Please running. Please running. Please continue. So we're going to talk about continue. So we're going to talk about continue. So we're going to talk about AI. And as Scott said, I'm Yina. I lead AI. And as Scott said, I'm Yina. I lead AI. And as Scott said, I'm Yina. I lead product for Azure AI and I'm here product for Azure AI and I'm here product for Azure AI and I'm here representing the work of many, many, representing the work of many, many, representing the work of many, many, many engineers, marketeers, data many engineers, marketeers, data many engineers, marketeers, data scientists at Microsoft who are bringing scientists at Microsoft who are bringing scientists at Microsoft who are bringing the platform for you. But what is Azure the platform for you. But what is Azure the platform for you. But what is Azure AI foundry? What is the open, flexible AI foundry? What is the open, flexible AI foundry? What is the open, flexible and secure platform that enables you as and secure platform that enables you as and secure platform that enables you as a developer to infuse AI in every single a developer to infuse AI in every single a developer to infuse AI in every single application that you do whether it is a application that you do whether it is a application that you do whether it is a brand new application or something that brand new application or something that brand new application or something that you've been working for a while you can you've been working for a while you can you've been working for a while you can use Azure AI foundry to like bring in use Azure AI foundry to like bring in use Azure AI foundry to like bring in all of the AI into your apps and as a all of the AI into your apps and as a all of the AI into your apps and as a developer it gives you and we're going developer it gives you and we're going developer it gives you and we're going to talk about that in detail about three to talk about that in detail about three to talk about that in detail about three core things one is models second one is core things one is models second one is core things one is models second one is an agentic platform and the Third is an agentic platform and the Third is an agentic platform and the Third is around all of the observability tools around all of the observability tools around all of the observability tools that you need as a as a developer to that you need as a as a developer to that you need as a as a developer to take your idea to code and your code to take your idea to code and your code to take your idea to code and your code to production. Now, we wanted to today in production. Now, we wanted to today in production. Now, we wanted to today in order to make the foundry real for you, order to make the foundry real for you, order to make the foundry real for you, we wanted to use an example of like a we wanted to use an example of like a we wanted to use an example of like a process that Scott does every single process that Scott does every single process that Scott does every single week. Yeah. As part of uh publishing week. Yeah. As part of uh publishing week. Yeah. As part of uh publishing handsel minutes and looking at like that handsel minutes and looking at like that handsel minutes and looking at like that end to end process and then Yeah. maybe

end to end process and then Yeah. maybe end to end process and then Yeah. maybe making it better with AI. So, I' I've making it better with AI. So, I' I've making it better with AI. So, I' I've got this podcast that I do on the side. got this podcast that I do on the side. got this podcast that I do on the side. Uh Microsoft has a very generous uh Uh Microsoft has a very generous uh Uh Microsoft has a very generous uh what's called a moonlighting if you do what's called a moonlighting if you do what's called a moonlighting if you do work on the side that's on your own work on the side that's on your own work on the side that's on your own computer in your own spare time. And computer in your own spare time. And computer in your own spare time. And I've been doing this podcast for the I've been doing this podcast for the I've been doing this podcast for the last 20 plus years. Uh it's something last 20 plus years. Uh it's something last 20 plus years. Uh it's something that I do on my own. Uh and I do it on that I do on my own. Uh and I do it on that I do on my own. Uh and I do it on my own money and then I pay a young lady my own money and then I pay a young lady my own money and then I pay a young lady named Mandy Moore to be my editor. So, named Mandy Moore to be my editor. So, named Mandy Moore to be my editor. So, it's just Mandy and I. And the show has it's just Mandy and I. And the show has it's just Mandy and I. And the show has I just published episode I just published episode I just published episode 997. So this is actually Thank 997. So this is actually Thank 997. So this is actually Thank you. This is this is what we call a a you. This is this is what we call a a you. This is this is what we call a a corpus a large corpus of material here. corpus a large corpus of material here. corpus a large corpus of material here. There's over 500 hours of material There's over 500 hours of material There's over 500 hours of material talking to cool people. In fact, I was talking to cool people. In fact, I was talking to cool people. In fact, I was actually uh almost delisted by a podcast actually uh almost delisted by a podcast actually uh almost delisted by a podcast directory when a young person sent an directory when a young person sent an directory when a young person sent an email saying that they believed that the email saying that they believed that the email saying that they believed that the podcast itself was AI podcast itself was AI podcast itself was AI generated because they couldn't conceive generated because they couldn't conceive generated because they couldn't conceive about that amount of work. Now, I feel about that amount of work. Now, I feel about that amount of work. Now, I feel very strongly about this work and as a very strongly about this work and as a very strongly about this work and as a human who created some art, I want to human who created some art, I want to human who created some art, I want to make sure that I'm going to be using AI make sure that I'm going to be using AI make sure that I'm going to be using AI in a way that only takes away the parts in a way that only takes away the parts in a way that only takes away the parts of the job that suck, like show notes.

of the job that suck, like show notes. of the job that suck, like show notes. And if you've ever listened to the show, And if you've ever listened to the show, And if you've ever listened to the show, I am notorious for saying, "I'm going to I am notorious for saying, "I'm going to I am notorious for saying, "I'm going to put that in the show notes, and then put that in the show notes, and then put that in the show notes, and then you'll never hear from me again." Never you'll never hear from me again." Never you'll never hear from me again." Never happens. Never happens. 10,000 episodes, happens. Never happens. 10,000 episodes, happens. Never happens. 10,000 episodes, close to 1,000 episodes. Yep. Hours and close to 1,000 episodes. Yep. Hours and close to 1,000 episodes. Yep. Hours and hours and hours of audio files, lots of hours and hours of audio files, lots of hours and hours of audio files, lots of broken promises. Lots of great people in broken promises. Lots of great people in broken promises. Lots of great people in the show, too, right? Yeah. Oh, and I the show, too, right? Yeah. Oh, and I the show, too, right? Yeah. Oh, and I have to look up their bios and who they have to look up their bios and who they have to look up their bios and who they are and if they have a Wikipedia page or are and if they have a Wikipedia page or are and if they have a Wikipedia page or something and it's a whole thing. And something and it's a whole thing. And something and it's a whole thing. And then they never email me back about then they never email me back about then they never email me back about their bios. There's a bunch of toil. their bios. There's a bunch of toil. their bios. There's a bunch of toil. There's figuring out the guests, There's figuring out the guests, There's figuring out the guests, scheduling the guests, packaging up the scheduling the guests, packaging up the scheduling the guests, packaging up the podcast. This is the part of the job podcast. This is the part of the job podcast. This is the part of the job that I don't like. And I think that that I don't like. And I think that that I don't like. And I think that computers should do toil. they should do computers should do toil. they should do computers should do toil. they should do work that is dirty, dull, or dangerous. work that is dirty, dull, or dangerous. work that is dirty, dull, or dangerous. The thing that you don't want to do, the The thing that you don't want to do, the The thing that you don't want to do, the things I don't want to do. So that's things I don't want to do. So that's things I don't want to do. So that's what we did as part of like showing all what we did as part of like showing all what we did as part of like showing all of the capabilities of Azure Foundry. We of the capabilities of Azure Foundry. We of the capabilities of Azure Foundry. We took this process that Scott has been took this process that Scott has been took this process that Scott has been doing for 20 years. The first step that doing for 20 years. The first step that doing for 20 years. The first step that we did is take those close to 1,000 we did is take those close to 1,000 we did is take those close to 1,000 audio files and process them with AI to audio files and process them with AI to audio files and process them with AI to get transcriptions and then use use the get transcriptions and then use use the get transcriptions and then use use the continue the process we're going to continue the process we're going to continue the process we're going to build through the throughout the show build through the throughout the show build through the throughout the show today to erase that toil at scale.

today to erase that toil at scale. today to erase that toil at scale. Right? And we wanted to do it in a way Right? And we wanted to do it in a way Right? And we wanted to do it in a way that put me in control because it's my that put me in control because it's my that put me in control because it's my show and I own this show and I don't show and I own this show and I don't show and I own this show and I don't want to just copy paste my entire show want to just copy paste my entire show want to just copy paste my entire show into some chatbot in the cloud. I want into some chatbot in the cloud. I want into some chatbot in the cloud. I want to know what's going on at every step. to know what's going on at every step. to know what's going on at every step. Now, there's a lot of things that we do Now, there's a lot of things that we do Now, there's a lot of things that we do here. There's a lot of pain here on the here. There's a lot of pain here on the here. There's a lot of pain here on the left that steals time. There's dealing left that steals time. There's dealing left that steals time. There's dealing with the show notes. There's links. If I with the show notes. There's links. If I with the show notes. There's links. If I have an AI generate the links, do I even have an AI generate the links, do I even have an AI generate the links, do I even know if they're real links? There's text know if they're real links? There's text know if they're real links? There's text summarization, which I'm constantly summarization, which I'm constantly summarization, which I'm constantly doing. And I'm spending probably two or doing. And I'm spending probably two or doing. And I'm spending probably two or three hours of my spare time, every three hours of my spare time, every three hours of my spare time, every single week for the last 20 years with administr. I want a human in the loop. administr. I want a human in the loop. Exactly. Okay, so that's the setup for Exactly. Okay, so that's the setup for Exactly. Okay, so that's the setup for what you're going to see us building what you're going to see us building what you're going to see us building today on stage and uh we're going to use today on stage and uh we're going to use today on stage and uh we're going to use that we're going to do the set the that we're going to do the set the that we're going to do the set the session in three chapters. We're going session in three chapters. We're going session in three chapters. We're going to first talk about models and then to first talk about models and then to first talk about models and then we're going to show all of the different we're going to show all of the different we're going to show all of the different set of models that we use for this set of models that we use for this set of models that we use for this process. We're then going to talk about process. We're then going to talk about process. We're then going to talk about agents and then we're going to show the agents and then we're going to show the agents and then we're going to show the different set of agents that we've different set of agents that we've different set of agents that we've built. And then last we're going to talk built. And then last we're going to talk built. And then last we're going to talk about observability and the set of tools about observability and the set of tools about observability and the set of tools that you can use to like as I mentioned that you can use to like as I mentioned that you can use to like as I mentioned before do all of the monitoring, before do all of the monitoring, before do all of the monitoring, tracing, logging that you need in order tracing, logging that you need in order tracing, logging that you need in order to take your applications to production.

to take your applications to production. to take your applications to production. So let's get going with models. So first So let's get going with models. So first So let's get going with models. So first of all, Azure Foundry has a lot of of all, Azure Foundry has a lot of of all, Azure Foundry has a lot of models, but that was not always the models, but that was not always the models, but that was not always the case. Like two years ago, the world was case. Like two years ago, the world was case. Like two years ago, the world was actually very simple. We had you know actually very simple. We had you know actually very simple. We had you know just a few foundational models but the just a few foundational models but the just a few foundational models but the rate range of change has been rate range of change has been rate range of change has been unprecedented over the last couple of unprecedented over the last couple of unprecedented over the last couple of years. We now have years. We now have years. We now have um tons of thousand models that are um tons of thousand models that are um tons of thousand models that are pushing the efficient frontier and pushing the efficient frontier and pushing the efficient frontier and creating new new choices or it's great creating new new choices or it's great creating new new choices or it's great opportunity for you as a developer opportunity for you as a developer opportunity for you as a developer because you have great capabilities and because you have great capabilities and because you have great capabilities and functionality but it also makes it quite functionality but it also makes it quite functionality but it also makes it quite complex to figure out what is the right complex to figure out what is the right complex to figure out what is the right model for the job. Well, in Azure I model for the job. Well, in Azure I model for the job. Well, in Azure I foundry, we have all of these different foundry, we have all of these different foundry, we have all of these different set of models and this at the conference set of models and this at the conference set of models and this at the conference we're announcing new set of capabilities we're announcing new set of capabilities we're announcing new set of capabilities on models. First, we're extending the on models. First, we're extending the on models. First, we're extending the set of models that we both host and sell set of models that we both host and sell set of models that we both host and sell in addition to the open AI family. We're in addition to the open AI family. We're in addition to the open AI family. We're bringing in deepseek mist meta black bringing in deepseek mist meta black bringing in deepseek mist meta black forest lab and uh um a set of models forest lab and uh um a set of models forest lab and uh um a set of models that not only provide that unified that not only provide that unified that not only provide that unified access but also enable you to like easy access but also enable you to like easy access but also enable you to like easy switch switch them from your code. Now switch switch them from your code. Now switch switch them from your code. Now I'm going to show how do you uh go into I'm going to show how do you uh go into I'm going to show how do you uh go into the product. So I'm going to switch to the product. So I'm going to switch to the product. So I'm going to switch to two. Here we are. Here this is the UI two. Here we are. Here this is the UI two. Here we are. Here this is the UI for Azure AI Foundry. We're going to for Azure AI Foundry. We're going to for Azure AI Foundry. We're going to right here in the model catalog. We're right here in the model catalog. We're right here in the model catalog. We're going to see the set of experiences that going to see the set of experiences that going to see the set of experiences that it offers for you. First of all is like it offers for you. First of all is like it offers for you. First of all is like a wide range of model selections. We a wide range of model selections. We a wide range of model selections. We have over 10,000 models now based on our have over 10,000 models now based on our have over 10,000 models now based on our far part partnership with Hagging Face.

far part partnership with Hagging Face. far part partnership with Hagging Face. We see here on the top the latest news We see here on the top the latest news We see here on the top the latest news of our announcements that we are making of our announcements that we are making of our announcements that we are making here today and every con every single here today and every con every single here today and every con every single day, every single week they're being day, every single week they're being day, every single week they're being updated. There's a set of models um all updated. There's a set of models um all updated. There's a set of models um all of the models that are listed here and of the models that are listed here and of the models that are listed here and there's a set of dimensions that you there's a set of dimensions that you there's a set of dimensions that you could use whether it is by model could use whether it is by model could use whether it is by model provider by industry by the set of provider by industry by the set of provider by industry by the set of capabilities whether they're are capabilities whether they're are capabilities whether they're are serverless or or ma available on manage serverless or or ma available on manage serverless or or ma available on manage compute the set of tasks that they compute the set of tasks that they compute the set of tasks that they support and all of those capabilities. support and all of those capabilities. support and all of those capabilities. But if you want to just ask for example But if you want to just ask for example But if you want to just ask for example we have a built-in agent into our portal we have a built-in agent into our portal we have a built-in agent into our portal where you can ask like what is the set where you can ask like what is the set where you can ask like what is the set of like models that uh for example if of like models that uh for example if of like models that uh for example if I'm doing like protein research so I can I'm doing like protein research so I can I'm doing like protein research so I can say best model for protein research say best model for protein research say best model for protein research do you have a model to talk to you about do you have a model to talk to you about do you have a model to talk to you about the models to pick Americ and uh I'm going to send that in and I'm and uh I'm going to send that in and I'm going to also show the leaderboards going to also show the leaderboards going to also show the leaderboards while that is going on um get some space while that is going on um get some space while that is going on um get some space leaderboards is the place where you want leaderboards is the place where you want leaderboards is the place where you want to come and see like what are the best to come and see like what are the best to come and see like what are the best models based on certain benchmarks. So models based on certain benchmarks. So models based on certain benchmarks. So we right now we have benchmarks around we right now we have benchmarks around we right now we have benchmarks around quality, safety, cost, throughput and quality, safety, cost, throughput and quality, safety, cost, throughput and you can see the set of uh trade-off you can see the set of uh trade-off you can see the set of uh trade-off charts. So if you want to select the charts. So if you want to select the charts. So if you want to select the best model depending on like a quality best model depending on like a quality best model depending on like a quality and cost comparison or a quality and and cost comparison or a quality and and cost comparison or a quality and throughput comparison and it will show throughput comparison and it will show throughput comparison and it will show you what is the best model depending on you what is the best model depending on you what is the best model depending on that scenario. Is are those leaderboards that scenario. Is are those leaderboards that scenario. Is are those leaderboards real? That's not like hardcoded, right?

real? That's not like hardcoded, right? real? That's not like hardcoded, right? No, this is not hardcoded. We're No, this is not hardcoded. We're No, this is not hardcoded. We're constantly like making doing benchmarks constantly like making doing benchmarks constantly like making doing benchmarks on the models and bringing those into on the models and bringing those into on the models and bringing those into the catalog. So if somebody's lower on the catalog. So if somebody's lower on the catalog. So if somebody's lower on the model or higher on the model that the model or higher on the model that the model or higher on the model that they're going to want to it's a they're going to want to it's a they're going to want to it's a competition to be better. Exactly. competition to be better. Exactly. competition to be better. Exactly. There's always a competition. Okay. Now There's always a competition. Okay. Now There's always a competition. Okay. Now you can also use your own data to you can also use your own data to you can also use your own data to evaluate models. So like you can uh evaluate models. So like you can uh evaluate models. So like you can uh bring in and uh use your existing data bring in and uh use your existing data bring in and uh use your existing data set if you have one or you can use an AI set if you have one or you can use an AI set if you have one or you can use an AI generated data set uh as generated data set uh as generated data set uh as well. Okay. There's so many people in well. Okay. There's so many people in well. Okay. There's so many people in the room. the room. the room. We're going to the other thing that we We're going to the other thing that we We're going to the other thing that we are announcing at build. Let's see if we are announcing at build. Let's see if we are announcing at build. Let's see if we can get the answer from the models. Yes. can get the answer from the models. Yes. can get the answer from the models. Yes. You see these are our best models for You see these are our best models for You see these are our best models for protein research is EVO uh diff and boom protein research is EVO uh diff and boom protein research is EVO uh diff and boom and tam genen. Those are models that are and tam genen. Those are models that are and tam genen. Those are models that are coming in from our foundry labs which is coming in from our foundry labs which is coming in from our foundry labs which is the way that we bring in all of the the way that we bring in all of the the way that we bring in all of the models from Microsoft research into the models from Microsoft research into the models from Microsoft research into the into the catalog. Now let's look at into the catalog. Now let's look at into the catalog. Now let's look at another announcement. This is model another announcement. This is model another announcement. This is model router. Model router takes the selection router. Model router takes the selection router. Model router takes the selection toil from your head. So it is a router toil from your head. So it is a router toil from your head. So it is a router on top of like the best models reasoning on top of like the best models reasoning on top of like the best models reasoning models the state-of-the-art models.

models the state-of-the-art models. models the state-of-the-art models. right now only on OpenAI models but soon right now only on OpenAI models but soon right now only on OpenAI models but soon across more models and here you can see across more models and here you can see across more models and here you can see I made it a simple question where is I made it a simple question where is I made it a simple question where is Hanoi and it used a nano model it's a Hanoi and it used a nano model it's a Hanoi and it used a nano model it's a very small model to answer that very very small model to answer that very very small model to answer that very simple question if I ask a more a little simple question if I ask a more a little simple question if I ask a more a little bit more complex question about planning bit more complex question about planning bit more complex question about planning a trip I have a set of constraints it's a trip I have a set of constraints it's a trip I have a set of constraints it's going to use uh in this case well it use going to use uh in this case well it use going to use uh in this case well it use a mini model which is like a better a mini model which is like a better a mini model which is like a better model for that job and if I ask a model for that job and if I ask a model for that job and if I ask a different question like for example what different question like for example what different question like for example what clothes should I bring um I don't want clothes should I bring um I don't want clothes should I bring um I don't want to go and waste money waste energy waste to go and waste money waste energy waste to go and waste money waste energy waste the environment on something like what's the environment on something like what's the environment on something like what's the capital of Hanoi exactly a nano the capital of Hanoi exactly a nano the capital of Hanoi exactly a nano model is it's going to use a small model model is it's going to use a small model model is it's going to use a small model again and if I give it a more again and if I give it a more again and if I give it a more complicated query then it's going to use complicated query then it's going to use complicated query then it's going to use something like a reasoning model for something like a reasoning model for something like a reasoning model for this now I mentioned that we took was this now I mentioned that we took was this now I mentioned that we took was close to 1,000 episodes right and then close to 1,000 episodes right and then close to 1,000 episodes right and then we process all of those we actually use we process all of those we actually use we process all of those we actually use uh whisper as a model to take the audio uh whisper as a model to take the audio uh whisper as a model to take the audio files and generate the transcriptions files and generate the transcriptions files and generate the transcriptions but then we wanted to like not just um but then we wanted to like not just um but then we wanted to like not just um we wanted to do a distillation of a we wanted to do a distillation of a we wanted to do a distillation of a model so that we can like reduce the model so that we can like reduce the model so that we can like reduce the cost that it takes to process the next cost that it takes to process the next cost that it takes to process the next 1,000 Right? Because Hanselman's 1,000 Right? Because Hanselman's 1,000 Right? Because Hanselman's international is me and Mandy. Like we international is me and Mandy. Like we international is me and Mandy. Like we don't have any money. So we would like don't have any money. So we would like don't have any money. So we would like to make our cost as low as possible.

to make our cost as low as possible. to make our cost as low as possible. Upfront cost is concerning to me, but Upfront cost is concerning to me, but Upfront cost is concerning to me, but also I got to do another thousand also I got to do another thousand also I got to do another thousand episodes over the next 20 years, right? episodes over the next 20 years, right? episodes over the next 20 years, right? I'd like that to be very like a penny. I'd like that to be very like a penny. I'd like that to be very like a penny. So it costs us about like a $100 to do So it costs us about like a $100 to do So it costs us about like a $100 to do the processing of the first 10,000 the processing of the first 10,000 the processing of the first 10,000 episodes and then like we do we did a episodes and then like we do we did a episodes and then like we do we did a fine-tuned model and the next 1,000 is fine-tuned model and the next 1,000 is fine-tuned model and the next 1,000 is going to cost us about $150. $150 going to cost us about $150. $150 going to cost us about $150. $150 because we're using a custom model because we're using a custom model because we're using a custom model smaller model faster and then you know smaller model faster and then you know smaller model faster and then you know it's customized specifically for your it's customized specifically for your it's customized specifically for your job and then like right now at uh build job and then like right now at uh build job and then like right now at uh build we are announcing developer uh tier that we are announcing developer uh tier that we are announcing developer uh tier that enables you to do experimentation uh enables you to do experimentation uh enables you to do experimentation uh reducing the costing hosting fees and reducing the costing hosting fees and reducing the costing hosting fees and then there is no SLA but like you can then there is no SLA but like you can then there is no SLA but like you can get to get to do all of the experiments get to get to do all of the experiments get to get to do all of the experiments on on on fine-tuning. Okay, we're also announcing fine-tuning. Okay, we're also announcing fine-tuning. Okay, we're also announcing great things around Foundry local. Oh great things around Foundry local. Oh great things around Foundry local. Oh yeah. So this is another thing that I yeah. So this is another thing that I yeah. So this is another thing that I feel that is really really cool is that feel that is really really cool is that feel that is really really cool is that I've got a machine here that is very I've got a machine here that is very I've got a machine here that is very capable. This is a laptop. This is a capable. This is a laptop. This is a capable. This is a laptop. This is a studio 2. I've got a 4060 here. I can do studio 2. I've got a 4060 here. I can do studio 2. I've got a 4060 here. I can do some of this work locally. So we've got some of this work locally. So we've got some of this work locally. So we've got Azure AI Foundry in the cloud. We've got Azure AI Foundry in the cloud. We've got Azure AI Foundry in the cloud. We've got Azure uh Foundry local here. I want to Azure uh Foundry local here. I want to Azure uh Foundry local here. I want to be able to look up people's be able to look up people's be able to look up people's um biographies like yourself. Now I um biographies like yourself. Now I um biographies like yourself. Now I could go out there and Google with Bing could go out there and Google with Bing could go out there and Google with Bing and then figure out your biography and and then figure out your biography and and then figure out your biography and then rewrite it for myself or copy paste then rewrite it for myself or copy paste then rewrite it for myself or copy paste it or I could use a combination of it or I could use a combination of it or I could use a combination of something like perplexity uh to go and something like perplexity uh to go and something like perplexity uh to go and figure out a lot of information about figure out a lot of information about figure out a lot of information about you and then summarize that and distill you and then summarize that and distill you and then summarize that and distill that down. So here we're going to start that down. So here we're going to start that down. So here we're going to start up Foundry local with the 54 mini model up Foundry local with the 54 mini model up Foundry local with the 54 mini model and we're going to start that on this

and we're going to start that on this and we're going to start that on this local service. So this is running on local service. So this is running on local service. So this is running on local host. Now you notice down here in local host. Now you notice down here in local host. Now you notice down here in the corner we just saw the GPU kick in the corner we just saw the GPU kick in the corner we just saw the GPU kick in and we just loaded FI4 mini into the and we just loaded FI4 mini into the and we just loaded FI4 mini into the local uh memory and we'll go and say local uh memory and we'll go and say local uh memory and we'll go and say ENA. Now this is going to come uh the ENA. Now this is going to come uh the ENA. Now this is going to come uh the first part is going to be a call to first part is going to be a call to first part is going to be a call to perplexity which is in the cloud to perplexity which is in the cloud to perplexity which is in the cloud to collect me a huge chunk of data and then collect me a huge chunk of data and then collect me a huge chunk of data and then we're going to go and run locally on the we're going to go and run locally on the we're going to go and run locally on the GPU and uh summarize and squish all of GPU and uh summarize and squish all of GPU and uh summarize and squish all of that. There we go. Now that's all that. There we go. Now that's all that. There we go. Now that's all running locally. Oh, now this is running locally. Oh, now this is running locally. Oh, now this is exciting because you were apparently Is exciting because you were apparently Is exciting because you were apparently Is that true? No, that's not true. It's It that true? No, that's not true. It's It that true? No, that's not true. It's It got confused as Mesel. You're fantastic. got confused as Mesel. You're fantastic. got confused as Mesel. You're fantastic. So, you're not Did I misspell your name? So, you're not Did I misspell your name? So, you're not Did I misspell your name? Did you? I think I probably spelled it Did you? I think I probably spelled it Did you? I think I probably spelled it wrong. No, let's do it again. Let's try. wrong. No, let's do it again. Let's try. wrong. No, let's do it again. Let's try. See, cuz AI can be wrong sometimes. See, cuz AI can be wrong sometimes. See, cuz AI can be wrong sometimes. I didn't realize that you were such a I didn't realize that you were such a I didn't realize that you were such a talent, a multi-talented. Uh, you were talent, a multi-talented. Uh, you were talent, a multi-talented. Uh, you were in Chicago. From Colombia. Young. Well, in Chicago. From Colombia. Young. Well, in Chicago. From Colombia. Young. Well, no, you're not. You're Australian. no, you're not. You're Australian. no, you're not. You're Australian. Oh, no. Hey, there you go. There you Oh, no. Hey, there you go. There you Oh, no. Hey, there you go. There you go.

go. go. So, you can see the little the little uh So, you can see the little the little uh So, you can see the little the little uh jump right here where the work happened jump right here where the work happened jump right here where the work happened locally on that machine. And then we'll locally on that machine. And then we'll locally on that machine. And then we'll go and we'll see uh what kind of musical go and we'll see uh what kind of musical go and we'll see uh what kind of musical theater background that I have uh when theater background that I have uh when theater background that I have uh when we run this as well. And you can notice we run this as well. And you can notice we run this as well. And you can notice the GPU memory as the GPU memory as the GPU memory as well. well. well. Cool. And I want to call out that uh it Cool. And I want to call out that uh it Cool. And I want to call out that uh it worked really really hard for you worked really really hard for you worked really really hard for you because you're awesome and apparently I because you're awesome and apparently I because you're awesome and apparently I am. uh did not require a lot of effort am. uh did not require a lot of effort am. uh did not require a lot of effort uh which is uh profoundly sad but then I uh which is uh profoundly sad but then I uh which is uh profoundly sad but then I can just go and say foundry service stop can just go and say foundry service stop can just go and say foundry service stop and then watch over here in the corner and then watch over here in the corner and then watch over here in the corner and then that GPU memory goes away so and then that GPU memory goes away so and then that GPU memory goes away so that's I can do that work locally all that's I can do that work locally all that's I can do that work locally all righty that's chapter one models you saw righty that's chapter one models you saw righty that's chapter one models you saw all of the demos of one we have packed all of the demos of one we have packed all of the demos of one we have packed sessions on models when going through sessions on models when going through sessions on models when going through the model catalog model router the the model catalog model router the the model catalog model router the fine-tuning capabilities foundry local fine-tuning capabilities foundry local fine-tuning capabilities foundry local Azure open AAI multimod model models, Azure open AAI multimod model models, Azure open AAI multimod model models, reasoning models. How do you optimize reasoning models. How do you optimize reasoning models. How do you optimize all of your Genai applications? Take the all of your Genai applications? Take the all of your Genai applications? Take the picture. Don't miss those sessions. Now, picture. Don't miss those sessions. Now, picture. Don't miss those sessions. Now, we're going to go to chapter two and we're going to go to chapter two and we're going to go to chapter two and we're going to talk all about we're going to talk all about we're going to talk all about agents. So, first of all, what's an agents. So, first of all, what's an agents. So, first of all, what's an agent for you, Scott? Well, so I'm agent for you, Scott? Well, so I'm agent for you, Scott? Well, so I'm confused about the difference between confused about the difference between confused about the difference between agents and tools because AIs don't have agents and tools because AIs don't have agents and tools because AIs don't have feet and they don't have hands and they feet and they don't have hands and they feet and they don't have hands and they can't do stuff, but sometimes you want can't do stuff, but sometimes you want can't do stuff, but sometimes you want an AI to like spin through some files.

an AI to like spin through some files. an AI to like spin through some files. So, I feel like that's a tool. That's a So, I feel like that's a tool. That's a So, I feel like that's a tool. That's a tool. I have been called a tool at work. tool. I have been called a tool at work. tool. I have been called a tool at work. Uh sometimes it happens. But then I feel Uh sometimes it happens. But then I feel Uh sometimes it happens. But then I feel like an agent might call be able to call like an agent might call be able to call like an agent might call be able to call multiple tools, but an agent like has a multiple tools, but an agent like has a multiple tools, but an agent like has a task where it's like take this and do it task where it's like take this and do it task where it's like take this and do it and then let me know when you're and then let me know when you're and then let me know when you're finished, right? I mean, we've been finished, right? I mean, we've been finished, right? I mean, we've been doing process automation for like ages, doing process automation for like ages, doing process automation for like ages, right? And the difference is now you're right? And the difference is now you're right? And the difference is now you're having an LLM that is helping you drive having an LLM that is helping you drive having an LLM that is helping you drive the control flow of the program. It's the control flow of the program. It's the control flow of the program. It's making some of those decisions for you. making some of those decisions for you. making some of those decisions for you. And it has a set of tools that they can And it has a set of tools that they can And it has a set of tools that they can invoke. Those can be retrieving invoke. Those can be retrieving invoke. Those can be retrieving information, grounding with knowledge. information, grounding with knowledge. information, grounding with knowledge. They can be actually acting like making They can be actually acting like making They can be actually acting like making an API request fold and then uh you know an API request fold and then uh you know an API request fold and then uh you know keeping a thread with memory of the keeping a thread with memory of the keeping a thread with memory of the current conversation, the current state. current conversation, the current state. current conversation, the current state. They can also support multimodel inputs They can also support multimodel inputs They can also support multimodel inputs like conversational images, video, text, like conversational images, video, text, like conversational images, video, text, speech and then can support a set of speech and then can support a set of speech and then can support a set of outputs that like you know let them and outputs that like you know let them and outputs that like you know let them and inputs and outputs that let them be inputs and outputs that let them be inputs and outputs that let them be invoked by other agents. Yeah. So one of invoked by other agents. Yeah. So one of invoked by other agents. Yeah. So one of the key things uh that we have is like the key things uh that we have is like the key things uh that we have is like the ability to orchestrate these agents.

the ability to orchestrate these agents. the ability to orchestrate these agents. I think that right now there's a lot of I think that right now there's a lot of I think that right now there's a lot of overload on when we say agents because overload on when we say agents because overload on when we say agents because like people then feel like they have to like people then feel like they have to like people then feel like they have to add a whole bunch of functionality to a add a whole bunch of functionality to a add a whole bunch of functionality to a single agent and what I usually like to single agent and what I usually like to single agent and what I usually like to um recommend is that instead of that um recommend is that instead of that um recommend is that instead of that like you create a smaller agent that has like you create a smaller agent that has like you create a smaller agent that has a specific set of task a specific a specific set of task a specific a specific set of task a specific functionality and then that you can functionality and then that you can functionality and then that you can validate and and evaluate just like unit validate and and evaluate just like unit validate and and evaluate just like unit testing and then like you can testing and then like you can testing and then like you can orchestrate them together to execute orchestrate them together to execute orchestrate them together to execute actually a process or a task and there actually a process or a task and there actually a process or a task and there we have two types of them that you will we have two types of them that you will we have two types of them that you will See as as you know as you go through the See as as you know as you go through the See as as you know as you go through the pro product the first one are connected pro product the first one are connected pro product the first one are connected agents which basically treat each other agents which basically treat each other agents which basically treat each other as tools. They invoke another agent just as tools. They invoke another agent just as tools. They invoke another agent just like an API request invocation gets a like an API request invocation gets a like an API request invocation gets a result and it continues with the test result and it continues with the test result and it continues with the test set of thing that they need to do. And set of thing that they need to do. And set of thing that they need to do. And the second one is more of a multi- aent the second one is more of a multi- aent the second one is more of a multi- aent workflow. Think about it. This is like workflow. Think about it. This is like workflow. Think about it. This is like where it's actually a human in the loop. where it's actually a human in the loop. where it's actually a human in the loop. It's a process that it needs to follow. It's a process that it needs to follow. It's a process that it needs to follow. It is not just letting the agents kind It is not just letting the agents kind It is not just letting the agents kind of do their thing, right? But actually of do their thing, right? But actually of do their thing, right? But actually like, you know, there's a process and like, you know, there's a process and like, you know, there's a process and that the second one is the one that that the second one is the one that that the second one is the one that we're going to be using today for the we're going to be using today for the we're going to be using today for the handsel minute podcast file, right?

handsel minute podcast file, right? handsel minute podcast file, right? Because I want that to fit into the way Because I want that to fit into the way Because I want that to fit into the way I do things now where I put my file into I do things now where I put my file into I do things now where I put my file into a folder in one drive. Mandy picks it a folder in one drive. Mandy picks it a folder in one drive. Mandy picks it up. She does the hard work of editing up. She does the hard work of editing up. She does the hard work of editing and cleaning it up. Then she will and cleaning it up. Then she will and cleaning it up. Then she will interact with an agent that's going to interact with an agent that's going to interact with an agent that's going to help us as a uh a team member that's help us as a uh a team member that's help us as a uh a team member that's helping us do the stuff that we don't helping us do the stuff that we don't helping us do the stuff that we don't want to do, like show notes. Then we'll want to do, like show notes. Then we'll want to do, like show notes. Then we'll validate the show notes, make sure that validate the show notes, make sure that validate the show notes, make sure that they're cool. So they're going to fit they're cool. So they're going to fit they're cool. So they're going to fit into our existing process of guest into our existing process of guest into our existing process of guest intake, packaging, and uh and promotion. intake, packaging, and uh and promotion. intake, packaging, and uh and promotion. So we're going to show a whole bunch of So we're going to show a whole bunch of So we're going to show a whole bunch of these different set of agents, pieces of these different set of agents, pieces of these different set of agents, pieces of them of the functionality that we've them of the functionality that we've them of the functionality that we've built, right? The voice orchestrator, we built, right? The voice orchestrator, we built, right? The voice orchestrator, we just saw biogenerator, the guest just saw biogenerator, the guest just saw biogenerator, the guest sourcing transcript, uh the show notes, sourcing transcript, uh the show notes, sourcing transcript, uh the show notes, and we're going to see different parts and we're going to see different parts and we're going to see different parts of the product as we go through this. of the product as we go through this. of the product as we go through this. Now, there's multiple ways in which you Now, there's multiple ways in which you Now, there's multiple ways in which you can build agents, and I think this is can build agents, and I think this is can build agents, and I think this is worth a conversation. Yeah, I would say worth a conversation. Yeah, I would say worth a conversation. Yeah, I would say that it is a little bit confusing that it is a little bit confusing that it is a little bit confusing because that word agent gets overloaded. because that word agent gets overloaded. because that word agent gets overloaded. If you want to feel smart at work, just If you want to feel smart at work, just If you want to feel smart at work, just say agentic in a meeting and people will say agentic in a meeting and people will say agentic in a meeting and people will think that you're fancy. Um, this is an think that you're fancy. Um, this is an think that you're fancy. Um, this is an agentic flow. Um, the way I've been thinking flow. Um, the way I've been thinking about it is when we started thinking about it is when we started thinking about it is when we started thinking about the cloud, right? You can do about the cloud, right? You can do about the cloud, right? You can do infrastructure as a service where you're infrastructure as a service where you're infrastructure as a service where you're buying a virtual machine and that's very buying a virtual machine and that's very buying a virtual machine and that's very low level. That's your infrastructure.

low level. That's your infrastructure. low level. That's your infrastructure. Then there's platform as a service where Then there's platform as a service where Then there's platform as a service where we're talking about Azure AI foundry we're talking about Azure AI foundry we're talking about Azure AI foundry where it feels like there's a lot of where it feels like there's a lot of where it feels like there's a lot of power but it's hiding complexity from power but it's hiding complexity from power but it's hiding complexity from me. I don't really know a managed me. I don't really know a managed me. I don't really know a managed service. So you don't have to think service. So you don't have to think service. So you don't have to think about your scaling your own GM your own about your scaling your own GM your own about your scaling your own GM your own GPUs. You don't have to think about like GPUs. You don't have to think about like GPUs. You don't have to think about like managing your entire fleet. You don't managing your entire fleet. You don't managing your entire fleet. You don't have to think about like geo residency have to think about like geo residency have to think about like geo residency like all of these different things you like all of these different things you like all of these different things you know exactly for you. And then software know exactly for you. And then software know exactly for you. And then software as a service would be more like co-pilot as a service would be more like co-pilot as a service would be more like co-pilot studio where I realistically have this studio where I realistically have this studio where I realistically have this runtime that I don't have to think about runtime that I don't have to think about runtime that I don't have to think about much of anything except a business much of anything except a business much of anything except a business problem. So we're sitting squarely at problem. So we're sitting squarely at problem. So we're sitting squarely at the platform as a service position which the platform as a service position which the platform as a service position which fits well with the way that I do stuff. fits well with the way that I do stuff. fits well with the way that I do stuff. And then at build we're announcing the And then at build we're announcing the And then at build we're announcing the GA general availability of the Azure AI GA general availability of the Azure AI GA general availability of the Azure AI foundry agent service which is that foundry agent service which is that foundry agent service which is that platform as a service uh service that platform as a service uh service that platform as a service uh service that enables you to like create agents enables you to like create agents enables you to like create agents declaratively and you know run them on declaratively and you know run them on declaratively and you know run them on the cloud use different set of models the cloud use different set of models the cloud use different set of models use different set of tools. The key use different set of tools. The key use different set of tools. The key difference from what we've seen being difference from what we've seen being difference from what we've seen being offered in in the past is the enterprise offered in in the past is the enterprise offered in in the past is the enterprise readiness and the capabilities that it readiness and the capabilities that it readiness and the capabilities that it has to like bring your own file storage has to like bring your own file storage has to like bring your own file storage to connect to your enterprise systems to to connect to your enterprise systems to to connect to your enterprise systems to support your organization's virtual support your organization's virtual support your organization's virtual network your uh you know you can bring network your uh you know you can bring network your uh you know you can bring your own um memory your own thread your own um memory your own thread your own um memory your own thread storage it supports authentication so storage it supports authentication so storage it supports authentication so like if you are have to authenticate like if you are have to authenticate like if you are have to authenticate with an API it supports passing all of with an API it supports passing all of with an API it supports passing all of the tokens as well as connecting with a the tokens as well as connecting with a the tokens as well as connecting with a set of tools like knowledge tools like set of tools like knowledge tools like set of tools like knowledge tools like fabric sharepoint in Bang for web fabric sharepoint in Bang for web fabric sharepoint in Bang for web grounding and Azure AI search. Now the

grounding and Azure AI search. Now the grounding and Azure AI search. Now the another key thing that I wanted to another key thing that I wanted to another key thing that I wanted to mention is how the foundry agent service mention is how the foundry agent service mention is how the foundry agent service is interoperable. So whether you're is interoperable. So whether you're is interoperable. So whether you're building the agent in our platform or building the agent in our platform or building the agent in our platform or you build an agent in another platform you build an agent in another platform you build an agent in another platform and you want to connect it with a and you want to connect it with a and you want to connect it with a foundry agent, we we support A2A, MCP, foundry agent, we we support A2A, MCP, foundry agent, we we support A2A, MCP, we're working with Langchen and Crew AI we're working with Langchen and Crew AI we're working with Langchen and Crew AI to support their uh languages as well in to support their uh languages as well in to support their uh languages as well in addition to supporting the open API the addition to supporting the open API the addition to supporting the open API the open AI uh assistance and responses API. open AI uh assistance and responses API. open AI uh assistance and responses API. Now let's see this in action. Let's show Now let's see this in action. Let's show Now let's see this in action. Let's show them some of the uh agents that we've them some of the uh agents that we've them some of the uh agents that we've built. Let me go to two. All right. So, built. Let me go to two. All right. So, built. Let me go to two. All right. So, we're switching over to your machine and we're switching over to your machine and we're switching over to your machine and uh I think show notes, right, is the one uh I think show notes, right, is the one uh I think show notes, right, is the one that is uh hurting me the most. So, that is uh hurting me the most. So, that is uh hurting me the most. So, we're here in agents. We're going to see we're here in agents. We're going to see we're here in agents. We're going to see the list of some of the agents that we the list of some of the agents that we the list of some of the agents that we created and we're going to show Yes, created and we're going to show Yes, created and we're going to show Yes, show notes. Let's show them show notes. show notes. Let's show them show notes. show notes. Let's show them show notes. Uh first thing that we're going to see Uh first thing that we're going to see Uh first thing that we're going to see here in show notes is that we're here in show notes is that we're here in show notes is that we're actually for this particular agent, actually for this particular agent, actually for this particular agent, we're using the nano model that we we're using the nano model that we we're using the nano model that we fine-tune in the past. Yeah. And I fine-tune in the past. Yeah. And I fine-tune in the past. Yeah. And I really want to call this out because really want to call this out because really want to call this out because this is important to me. I understand this is important to me. I understand this is important to me. I understand that yes, I could upload my uh MP3 into that yes, I could upload my uh MP3 into that yes, I could upload my uh MP3 into a giant chatbot. Like that's a thing you a giant chatbot. Like that's a thing you a giant chatbot. Like that's a thing you could do, but that is really bringing a could do, but that is really bringing a could do, but that is really bringing a sledgehammer to a scalpel fight. It's sledgehammer to a scalpel fight. It's sledgehammer to a scalpel fight. It's going to cost me way more money. It's going to cost me way more money. It's going to cost me way more money. It's going to use way more power. It's going to use way more power. It's going to use way more power. It's completely unnecessary. And if I got to completely unnecessary. And if I got to completely unnecessary. And if I got to do it a thousand times, it's going to be do it a thousand times, it's going to be do it a thousand times, it's going to be a problem because I want to spend as a problem because I want to spend as a problem because I want to spend as little money as as possible. This is a little money as as possible. This is a little money as as possible. This is a is it distilled is the word distilled is it distilled is the word distilled is it distilled is the word distilled model that is a tiny model the smallest model that is a tiny model the smallest model that is a tiny model the smallest possible model that doesn't give me possible model that doesn't give me possible model that doesn't give me relationship advice. It doesn't tell me relationship advice. It doesn't tell me relationship advice. It doesn't tell me how tall Brad Pitt is. It is trained on

how tall Brad Pitt is. It is trained on how tall Brad Pitt is. It is trained on how to do show notes and that's how to do show notes and that's how to do show notes and that's important. It does the exact thing I ask important. It does the exact thing I ask important. It does the exact thing I ask it to do and no more. So we're going to it to do and no more. So we're going to it to do and no more. So we're going to paste the transcription of one of your paste the transcription of one of your paste the transcription of one of your shows. Okay. And then we're gonna see shows. Okay. And then we're gonna see shows. Okay. And then we're gonna see which which show did you do? This is which which show did you do? This is which which show did you do? This is actually a show with one of my actually a show with one of my actually a show with one of my co-workers and dear friends Raji show uh co-workers and dear friends Raji show uh co-workers and dear friends Raji show uh 881 Raji Raj Gopalan we did a talk about 881 Raji Raj Gopalan we did a talk about 881 Raji Raj Gopalan we did a talk about her book exactly so we can see here it's her book exactly so we can see here it's her book exactly so we can see here it's generating the show notes it has all of generating the show notes it has all of generating the show notes it has all of the different uh takeaways and notes the different uh takeaways and notes the different uh takeaways and notes hang on notable quotes did I say hang on notable quotes did I say hang on notable quotes did I say something pathy no I think you said something pathy no I think you said something pathy no I think you said something very smart things happened something very smart things happened something very smart things happened there zoom in Imposttor syndrome means there zoom in Imposttor syndrome means there zoom in Imposttor syndrome means you're doing something okay that's good you're doing something okay that's good you're doing something okay that's good thoughts are not facts that's deep I thoughts are not facts that's deep I thoughts are not facts that's deep I think that's Rajie, not you. Oh, that think that's Rajie, not you. Oh, that think that's Rajie, not you. Oh, that was probably not was probably not was probably not me. Okay, there's a couple of cool me. Okay, there's a couple of cool me. Okay, there's a couple of cool things that I want you all to see here. things that I want you all to see here. things that I want you all to see here. View code. It opens up a view of the View code. It opens up a view of the View code. It opens up a view of the this particular interaction that we're this particular interaction that we're this particular interaction that we're having. You can open in VS Code. It will having. You can open in VS Code. It will having. You can open in VS Code. It will open up a sample uh project for you open up a sample uh project for you open up a sample uh project for you right there in the browser. It has right there in the browser. It has right there in the browser. It has Python and C#. I want to make sure shout Python and C#. I want to make sure shout Python and C#. I want to make sure shout out to Sharp. You know I you he was very out to Sharp. You know I you he was very out to Sharp. You know I you he was very happy when he saw she make sure happy when he saw she make sure happy when he saw she make sure and then the other thing that I want to and then the other thing that I want to and then the other thing that I want to show is agent catalog. So here in case show is agent catalog. So here in case show is agent catalog. So here in case you want to get started with a different you want to get started with a different you want to get started with a different set of agents we have Microsoft build set of agents we have Microsoft build set of agents we have Microsoft build templates. We have also several from our templates. We have also several from our templates. We have also several from our partners. You saw in the keynote the one partners. You saw in the keynote the one partners. You saw in the keynote the one that was coming from Stanford uh on the that was coming from Stanford uh on the that was coming from Stanford uh on the healthcare example. So we have lots of healthcare example. So we have lots of healthcare example. So we have lots of different set of agents for you to get

different set of agents for you to get different set of agents for you to get started. But one of the things that we started. But one of the things that we started. But one of the things that we did and I'm going to go back to our did and I'm going to go back to our did and I'm going to go back to our agents here. of my agents was an agents here. of my agents was an agents here. of my agents was an orchestrator voice agent. Yeah, you saw orchestrator voice agent. Yeah, you saw orchestrator voice agent. Yeah, you saw me running things from the command line. me running things from the command line. me running things from the command line. I can do that. I could run a Python app, I can do that. I could run a Python app, I can do that. I could run a Python app, but I want Mandy on my team to be able but I want Mandy on my team to be able but I want Mandy on my team to be able to interact with this in a way that's to interact with this in a way that's to interact with this in a way that's comfortable. And we thought it'd be cool comfortable. And we thought it'd be cool comfortable. And we thought it'd be cool to have the orchestrator router agent to have the orchestrator router agent to have the orchestrator router agent that has all of the Hanselman available that has all of the Hanselman available that has all of the Hanselman available things to be something we could talk to, things to be something we could talk to, things to be something we could talk to, which would be cool. Well, let's try it. which would be cool. Well, let's try it. which would be cool. Well, let's try it. All right. Hey, are you there? And can you right. Hey, are you there? And can you give me a list of the things that you give me a list of the things that you give me a list of the things that you can do to help me with the podcast can do to help me with the podcast can do to help me with the podcast production? Of course, Scott. I'm here to help Of course, Scott. I'm here to help streamline and enhance your podcast streamline and enhance your podcast streamline and enhance your podcast production. Here's a list of what I can production. Here's a list of what I can production. Here's a list of what I can do for you. Guest intake. Collaborate do for you. Guest intake. Collaborate do for you. Guest intake. Collaborate with the guest intake agent to with the guest intake agent to with the guest intake agent to brainstorm and refine a list of brainstorm and refine a list of brainstorm and refine a list of potential guests to invite on your show. potential guests to invite on your show. potential guests to invite on your show. Help you create personalized outreach Help you create personalized outreach Help you create personalized outreach strategies to approach notable guests in strategies to approach notable guests in strategies to approach notable guests in industry, society, and technology.

industry, society, and technology. industry, society, and technology. Speaker bio creation. Use the Speaker bio creation. Use the Speaker bio creation. Use the biogenerator to create a draft speaker. biogenerator to create a draft speaker. biogenerator to create a draft speaker. I'm gonna go ahead and respectfully I'm gonna go ahead and respectfully I'm gonna go ahead and respectfully interrupt you. Uh, hang in there for a interrupt you. Uh, hang in there for a interrupt you. Uh, hang in there for a second there, bud. Okay, cool. Awesome. second there, bud. Okay, cool. Awesome. second there, bud. Okay, cool. Awesome. So, that's one of our agents. And then So, that's one of our agents. And then So, that's one of our agents. And then the next one we did was around search. the next one we did was around search. the next one we did was around search. Yeah. So, this was something I thought Yeah. So, this was something I thought Yeah. So, this was something I thought was really exciting. Let's switch over was really exciting. Let's switch over was really exciting. Let's switch over to my machine here. Remember, if it's my to my machine here. Remember, if it's my to my machine here. Remember, if it's my data and I'm running it in my Azure AI data and I'm running it in my Azure AI data and I'm running it in my Azure AI Foundry uh experience. So the model that Foundry uh experience. So the model that Foundry uh experience. So the model that we went and distilled is mine and we can we went and distilled is mine and we can we went and distilled is mine and we can ask it questions now. And this was this ask it questions now. And this was this ask it questions now. And this was this was freaking amazing because my search was freaking amazing because my search was freaking amazing because my search right now on my site is basically I just right now on my site is basically I just right now on my site is basically I just load a giant JSON file and I can search load a giant JSON file and I can search load a giant JSON file and I can search for people's names. Uh now I can chat for people's names. Uh now I can chat for people's names. Uh now I can chat with 20 years of stuff and I can say is with 20 years of stuff and I can say is with 20 years of stuff and I can say is Scott nice? Scott nice? Scott nice? And it's going to go and give me an AI And it's going to go and give me an AI And it's going to go and give me an AI generated proof. Yes. That I'm that generated proof. Yes. That I'm that generated proof. Yes. That I'm that you're actually nice. Well, I they might you're actually nice. Well, I they might you're actually nice. Well, I they might say, "Well, I don't know. He's probably say, "Well, I don't know. He's probably say, "Well, I don't know. He's probably not very nice. He's So, this is using not very nice. He's So, this is using not very nice. He's So, this is using Azure AI search with uh but we've looks.

Azure AI search with uh but we've looks. Azure AI search with uh but we've looks. Let me see. Look, it's got proof. Look, Let me see. Look, it's got proof. Look, Let me see. Look, it's got proof. Look, it's it's it's grounded. It's grounded. it's it's it's grounded. It's grounded. it's it's it's grounded. It's grounded. I was nice twice in a thousand I was nice twice in a thousand I was nice twice in a thousand episodes." episodes." episodes." Well, that's an easy that's an easy Well, that's an easy that's an easy Well, that's an easy that's an easy question, right? Like, what are the question, right? Like, what are the question, right? Like, what are the characteristics of a person named Scott? characteristics of a person named Scott? characteristics of a person named Scott? Well, I could have been any Scott at Well, I could have been any Scott at Well, I could have been any Scott at this point. So, no, this here is a more this point. So, no, this here is a more this point. So, no, this here is a more question. Yeah, let's ask a question. Yeah, let's ask a question. Yeah, let's ask a sophisticated question because this is sophisticated question because this is sophisticated question because this is more than just rag, right? Because rag more than just rag, right? Because rag more than just rag, right? Because rag is this this augmented retrieval over is this this augmented retrieval over is this this augmented retrieval over data. But what if we asked a complicated data. But what if we asked a complicated data. But what if we asked a complicated series of questions like you know uh series of questions like you know uh series of questions like you know uh over the last 20 years are there any over the last 20 years are there any over the last 20 years are there any recurring ideas or universal truths in recurring ideas or universal truths in recurring ideas or universal truths in the last thousand episodes that have the last thousand episodes that have the last thousand episodes that have emerged? So this is not just like a one emerged? So this is not just like a one emerged? So this is not just like a one shot, right? Like this is what happens shot, right? Like this is what happens shot, right? Like this is what happens with rag that we've been doing, right? with rag that we've been doing, right? with rag that we've been doing, right? It's like the going from 10 links to It's like the going from 10 links to It's like the going from 10 links to just like you get one shot to get the just like you get one shot to get the just like you get one shot to get the answer right from your index. Well, this answer right from your index. Well, this answer right from your index. Well, this uh one of the things that we're uh one of the things that we're uh one of the things that we're announcing it's dropping in, right? Like announcing it's dropping in, right? Like announcing it's dropping in, right? Like I see it like all of these different set I see it like all of these different set I see it like all of these different set of uh so let's see. Yeah, it's done.

of uh so let's see. Yeah, it's done. of uh so let's see. Yeah, it's done. Let's go look at this. So down here, Let's go look at this. So down here, Let's go look at this. So down here, look, it's split it up into multiple look, it's split it up into multiple look, it's split it up into multiple questions. spent time on each individual questions. spent time on each individual questions. spent time on each individual question, broke the subqueries up, and question, broke the subqueries up, and question, broke the subqueries up, and those subqueries are not subsentes. those subqueries are not subsentes. those subqueries are not subsentes. Those are they're subqueries. Then it Those are they're subqueries. Then it Those are they're subqueries. Then it goes and retrieves the answer for those goes and retrieves the answer for those goes and retrieves the answer for those queries and bring brings them back into queries and bring brings them back into queries and bring brings them back into the context of the model so that it can the context of the model so that it can the context of the model so that it can give you the answer. Pragmatism, give you the answer. Pragmatism, give you the answer. Pragmatism, mentorship, community. Cool. And so this mentorship, community. Cool. And so this mentorship, community. Cool. And so this is what we call aentic retrieval on is what we call aentic retrieval on is what we call aentic retrieval on Azure AI search. Yeah, that's really Azure AI search. Yeah, that's really Azure AI search. Yeah, that's really cool. I'm going to be able to ask a lot cool. I'm going to be able to ask a lot cool. I'm going to be able to ask a lot of interesting questions. Okay, let's go of interesting questions. Okay, let's go of interesting questions. Okay, let's go to the next one, which is how we did the to the next one, which is how we did the to the next one, which is how we did the transcripts. transcripts. transcripts. Oh yeah yeah yeah yeah yeah. So you can Oh yeah yeah yeah yeah yeah. So you can Oh yeah yeah yeah yeah yeah. So you can go in here and you can define a schema. go in here and you can define a schema. go in here and you can define a schema. This is if you don't have a thousand of This is if you don't have a thousand of This is if you don't have a thousand of something right? Let's say that you want something right? Let's say that you want something right? Let's say that you want that kind of feature but you don't have that kind of feature but you don't have that kind of feature but you don't have a large corpus to uh so we define we a large corpus to uh so we define we a large corpus to uh so we define we defined the schema which we're using defined the schema which we're using defined the schema which we're using content understanding which is a service content understanding which is a service content understanding which is a service in Azure foundry and then we process the in Azure foundry and then we process the in Azure foundry and then we process the same podcast with our friend Raji and same podcast with our friend Raji and same podcast with our friend Raji and then we created all of the different set then we created all of the different set then we created all of the different set of field results the title the summary of field results the title the summary of field results the title the summary and uh some of the uh was it the quotes and uh some of the uh was it the quotes and uh some of the uh was it the quotes here as well uh no we did uh topics here as well uh no we did uh topics here as well uh no we did uh topics topics topic summary and title and then topics topic summary and title and then topics topic summary and title and then we we we have here It's identifying uh we we we have here It's identifying uh we we we have here It's identifying uh speaker one, speaker two. It turns out speaker one, speaker two. It turns out speaker one, speaker two. It turns out it identified me as speakers two and it identified me as speakers two and it identified me as speakers two and three because apparently my three because apparently my three because apparently my advertisement voice is different from my advertisement voice is different from my advertisement voice is different from my actual normal voice. So when I did the actual normal voice. So when I did the actual normal voice. So when I did the ad, it thought that was a different guy, ad, it thought that was a different guy, ad, it thought that was a different guy, which I thought was crazy. Yeah. So you which I thought was crazy. Yeah. So you which I thought was crazy. Yeah. So you can see here you got all of this is can see here you got all of this is can see here you got all of this is Rajie speaking here and then here's me

Rajie speaking here and then here's me Rajie speaking here and then here's me jumping in. So it's identifying jumping in. So it's identifying jumping in. So it's identifying different speakers and then it's pulled different speakers and then it's pulled different speakers and then it's pulled out topics about mindfulness and out topics about mindfulness and out topics about mindfulness and sustainable habits and all that kind of sustainable habits and all that kind of sustainable habits and all that kind of stuff. So that's built in. That's stuff. So that's built in. That's stuff. So that's built in. That's another one of the agents that we got another one of the agents that we got another one of the agents that we got built connecting ash content built connecting ash content built connecting ash content understanding with an agent in AI understanding with an agent in AI understanding with an agent in AI foundry. Now once we got all of those foundry. Now once we got all of those foundry. Now once we got all of those agents, did we miss showing one? No, I agents, did we miss showing one? No, I agents, did we miss showing one? No, I think we're good. Once once we got all think we're good. Once once we got all think we're good. Once once we got all of those agents, we wanted to wire them of those agents, we wanted to wire them of those agents, we wanted to wire them up together in a multi- aent process. up together in a multi- aent process. up together in a multi- aent process. And for that, we're using semantic And for that, we're using semantic And for that, we're using semantic kernel and autoen. And we have we kernel and autoen. And we have we kernel and autoen. And we have we actually have a lot of time right now. actually have a lot of time right now. actually have a lot of time right now. So let's make sure to get dig in and So let's make sure to get dig in and So let's make sure to get dig in and look at some of this code because I want look at some of this code because I want look at some of this code because I want to make sure that we're showing real to make sure that we're showing real to make sure that we're showing real code because everything we built here is code because everything we built here is code because everything we built here is is legit. So this is C. Before Before is legit. So this is C. Before Before is legit. So this is C. Before Before you show that, let me make a couple of you show that, let me make a couple of you show that, let me make a couple of comments on semantic kernel. So we've comments on semantic kernel. So we've comments on semantic kernel. So we've been working on agentic frameworks for been working on agentic frameworks for been working on agentic frameworks for quite a while. We released autogen and quite a while. We released autogen and quite a while. We released autogen and semantic kernel in 2023 and there semantic kernel in 2023 and there semantic kernel in 2023 and there autogen has been coming in as an agentic autogen has been coming in as an agentic autogen has been coming in as an agentic framework from our Microsoft research framework from our Microsoft research framework from our Microsoft research organization. Semantic kernel is the the organization. Semantic kernel is the the organization. Semantic kernel is the the one that we've been recommending for one that we've been recommending for one that we've been recommending for using in production as like we bring all using in production as like we bring all using in production as like we bring all of that innovation into semantic kernel of that innovation into semantic kernel of that innovation into semantic kernel where now we're bringing them together where now we're bringing them together where now we're bringing them together into like one agentic framework and at into like one agentic framework and at into like one agentic framework and at build we're announcing like you know build we're announcing like you know build we're announcing like you know they're using the same runtime and then they're using the same runtime and then they're using the same runtime and then we're going to converge them into a we're going to converge them into a we're going to converge them into a single agentic framework as we go single agentic framework as we go single agentic framework as we go forward. Now here we're using semantic forward. Now here we're using semantic forward. Now here we're using semantic kernel to bring all of these agents kernel to bring all of these agents kernel to bring all of these agents together, wire them up and which we have together, wire them up and which we have together, wire them up and which we have the yes semantic kernel y library and we

the yes semantic kernel y library and we the yes semantic kernel y library and we have all of the different agents that have all of the different agents that have all of the different agents that we're using. Yeah. So over here we've we're using. Yeah. So over here we've we're using. Yeah. So over here we've got our content agent which is going to got our content agent which is going to got our content agent which is going to go and uh do our show notes. We've got go and uh do our show notes. We've got go and uh do our show notes. We've got link verification because we want to link verification because we want to link verification because we want to make sure that this is grounded in make sure that this is grounded in make sure that this is grounded in reality. when it runs through the show reality. when it runs through the show reality. when it runs through the show notes, oftentimes I'll say things like, notes, oftentimes I'll say things like, notes, oftentimes I'll say things like, "Well, that's great, Yina. I'll be sure "Well, that's great, Yina. I'll be sure "Well, that's great, Yina. I'll be sure to put a link to your book in the show to put a link to your book in the show to put a link to your book in the show notes and then you'll never hear from me notes and then you'll never hear from me notes and then you'll never hear from me again." We're recording right now. Well, again." We're recording right now. Well, again." We're recording right now. Well, let's ask, what do you want to put in let's ask, what do you want to put in let's ask, what do you want to put in the show notes? We want to put in the the show notes? We want to put in the the show notes? We want to put in the show notes one, the reference for the show notes one, the reference for the show notes one, the reference for the session for session for session for BRK15, which is our build talk, right? BRK15, which is our build talk, right? BRK15, which is our build talk, right? We want to put uh the link to the AI We want to put uh the link to the AI We want to put uh the link to the AI foundry website, which is aasher.com, foundry website, which is aasher.com, foundry website, which is aasher.com, right? But let's give it a little bit right? But let's give it a little bit right? But let's give it a little bit more complex. I think that you should um more complex. I think that you should um more complex. I think that you should um put a link to my uh my YouTube channel. put a link to my uh my YouTube channel. put a link to my uh my YouTube channel. I'm Scott and I'd like you to put a link I'm Scott and I'd like you to put a link I'm Scott and I'd like you to put a link to my YouTube. Oh, and then put a link to my YouTube. Oh, and then put a link to my YouTube. Oh, and then put a link to uh Hansel Minutes episode 881 with to uh Hansel Minutes episode 881 with to uh Hansel Minutes episode 881 with Rajie as a call back link. Yes. Yeah. Rajie as a call back link. Yes. Yeah. Rajie as a call back link. Yes. Yeah. So, make sure you do that, please. So, make sure you do that, please. So, make sure you do that, please. That'd be cool. That'd be cool. Yeah.

That'd be cool. That'd be cool. Yeah. That'd be cool. That'd be cool. Yeah. Yeah. Yeah. Okay. So, this application Yeah. Yeah. Okay. So, this application Yeah. Yeah. Okay. So, this application here has these these agents identified. here has these these agents identified. here has these these agents identified. And if we go and take a look at them, And if we go and take a look at them, And if we go and take a look at them, they are listed as semantic kernel they are listed as semantic kernel they are listed as semantic kernel functions, right? And then semantic functions, right? And then semantic functions, right? And then semantic kernel includes the ability to call an kernel includes the ability to call an kernel includes the ability to call an agent from it and participate in these agent from it and participate in these agent from it and participate in these kind of kind of kind of workflows and then from them we'll go workflows and then from them we'll go workflows and then from them we'll go and then just spin out the result and and then just spin out the result and and then just spin out the result and then the result will end up being in a then the result will end up being in a then the result will end up being in a markdown file called show notes. Okay, markdown file called show notes. Okay, markdown file called show notes. Okay, perfect. All right. Where are we here? Okay. Oh, and right. Where are we here? Okay. Oh, and then I wanted to show the YAML. Oh, and then I wanted to show the YAML. Oh, and then I wanted to show the YAML. Oh, and the and the mermaid chart. Oh, the the and the mermaid chart. Oh, the the and the mermaid chart. Oh, the mermaid chart is good. Okay. So we all mermaid chart is good. Okay. So we all mermaid chart is good. Okay. So we all know that I don't like know that I don't like know that I don't like YAML. So we click here and I don't have YAML. So we click here and I don't have YAML. So we click here and I don't have to look at to look at to look at YAML. YAML. YAML. Uh so this is a mermaid chart. Mermaid Uh so this is a mermaid chart. Mermaid Uh so this is a mermaid chart. Mermaid is great. That is showing me basically is great. That is showing me basically is great. That is showing me basically the results of that YAML showing that the results of that YAML showing that the results of that YAML showing that okay we're going to go and we're going okay we're going to go and we're going okay we're going to go and we're going to do the transcript. We're going to to do the transcript. We're going to to do the transcript. We're going to analyze the content for the show. analyze the content for the show. analyze the content for the show. Generate the show notes. We're going to Generate the show notes. We're going to Generate the show notes. We're going to go and verify each one of those links to go and verify each one of those links to go and verify each one of those links to make sure that they are legit. And then make sure that they are legit. And then make sure that they are legit. And then we're going to summarize the result and we're going to summarize the result and we're going to summarize the result and that'll be the complete the complete that'll be the complete the complete that'll be the complete the complete thing. That's awesome. So, that's pretty thing. That's awesome. So, that's pretty thing. That's awesome. So, that's pretty cool. So, all of that will will happen cool. So, all of that will will happen cool. So, all of that will will happen here.

here. here. While you're at it, one of the things While you're at it, one of the things While you're at it, one of the things that I forgot to mention, we're kind of that I forgot to mention, we're kind of that I forgot to mention, we're kind of speeding up because we're like we have speeding up because we're like we have speeding up because we're like we have so much. We're very excited, but we have so much. We're very excited, but we have so much. We're very excited, but we have we have a lot of content. you know on we have a lot of content. you know on we have a lot of content. you know on model router our current analysis tell model router our current analysis tell model router our current analysis tell us that like when you're using compare us that like when you're using compare us that like when you're using compare it to GPT41 and you're using model it to GPT41 and you're using model it to GPT41 and you're using model router you can get up to 60% price like router you can get up to 60% price like router you can get up to 60% price like uh off in the price meaning like more uh off in the price meaning like more uh off in the price meaning like more efficient less expense like what am I efficient less expense like what am I efficient less expense like what am I saying it's less expensive well it's got saying it's less expensive well it's got saying it's less expensive well it's got a less environmental hit it's got it's a less environmental hit it's got it's a less environmental hit it's got it's more efficient it's smaller like the more efficient it's smaller like the more efficient it's smaller like the idea is you put the router in front of idea is you put the router in front of idea is you put the router in front of it and it does it's a it's a load it and it does it's a it's a load it and it does it's a it's a load balancer that's the word that I was balancer that's the word that I was balancer that's the word that I was trying to find is cheaper it's 60% % trying to find is cheaper it's 60% % trying to find is cheaper it's 60% % cheaper. Okay. And then you only get one cheaper. Okay. And then you only get one cheaper. Okay. And then you only get one to two points hit in accuracy. Okay. to two points hit in accuracy. Okay. to two points hit in accuracy. Okay. Which is pretty significant. Yeah. Now, Which is pretty significant. Yeah. Now, Which is pretty significant. Yeah. Now, let me help me understand what percent let me help me understand what percent let me help me understand what percent of accuracy turns you into an Australian of accuracy turns you into an Australian of accuracy turns you into an Australian singer. I don't know. I think that's a singer. I don't know. I think that's a singer. I don't know. I think that's a very low one. very low one. very low one. Okay. Now, let's show um do we want to Okay. Now, let's show um do we want to Okay. Now, let's show um do we want to stop our podcast? Are we finished with stop our podcast? Are we finished with stop our podcast? Are we finished with our show?

our show? our show? What else do we want to say? We want to What else do we want to say? We want to What else do we want to say? We want to say we talked about models. Okay. Okay, say we talked about models. Okay. Okay, say we talked about models. Okay. Okay, we talked about modules digging into we talked about modules digging into we talked about modules digging into agents. Talked about semantic. What are agents. Talked about semantic. What are agents. Talked about semantic. What are the key things that we said about the key things that we said about the key things that we said about models? Well, there's a there's a metric models? Well, there's a there's a metric models? Well, there's a there's a metric ton of them, 11,000 of them. Yeah, ton of them, 11,000 of them. Yeah, ton of them, 11,000 of them. Yeah, there's models that are specific, not there's models that are specific, not there's models that are specific, not just large language models, but there just large language models, but there just large language models, but there are models of all different flavors, are models of all different flavors, are models of all different flavors, including specific models for things including specific models for things including specific models for things like protein, healthare, finance, yeah, like protein, healthare, finance, yeah, like protein, healthare, finance, yeah, all those things. You have um I think all those things. You have um I think all those things. You have um I think you saw I thought 11,000 models in you saw I thought 11,000 models in you saw I thought 11,000 models in there. All of those models can then be there. All of those models can then be there. All of those models can then be fine-tuned. And then one of the things fine-tuned. And then one of the things fine-tuned. And then one of the things that I was interested in that I didn't that I was interested in that I didn't that I was interested in that I didn't understand was uh you showed a graph understand was uh you showed a graph understand was uh you showed a graph like this on their fine-tuning, but I like this on their fine-tuning, but I like this on their fine-tuning, but I don't understand how you know that it's don't understand how you know that it's don't understand how you know that it's correct or not. Oh, and that one we correct or not. Oh, and that one we correct or not. Oh, and that one we specifically didn't do an evaluation specifically didn't do an evaluation specifically didn't do an evaluation run, but like you want to get the loss run, but like you want to get the loss run, but like you want to get the loss to zero and the accuracy closer to one. to zero and the accuracy closer to one. to zero and the accuracy closer to one. Can you can you show me it not looking Can you can you show me it not looking Can you can you show me it not looking good? Because I think it's always cool good? Because I think it's always cool good? Because I think it's always cool when we do demos, but show me a failure. when we do demos, but show me a failure. when we do demos, but show me a failure. Oh, we got some failures. Yes. When we Oh, we got some failures. Yes. When we Oh, we got some failures. Yes. When we were doing the runs, we have Nobody were doing the runs, we have Nobody were doing the runs, we have Nobody likes to see a demo that's like works. likes to see a demo that's like works. likes to see a demo that's like works. This one actually didn't go through. So This one actually didn't go through. So This one actually didn't go through. So this looks here that that looks like I'm this looks here that that looks like I'm this looks here that that looks like I'm not a scientist, but that doesn't look not a scientist, but that doesn't look not a scientist, but that doesn't look like a good line. Yeah, exactly. The like a good line. Yeah, exactly. The like a good line. Yeah, exactly. The line was going heading towards the line was going heading towards the line was going heading towards the middle and not necessarily trending middle and not necessarily trending middle and not necessarily trending towards one. So like So this is saying towards one. So like So this is saying towards one. So like So this is saying whether or not it generated show notes whether or not it generated show notes whether or not it generated show notes correctly.

correctly. correctly. And what we do is we verify that against And what we do is we verify that against And what we do is we verify that against actual show notes and actual actual show notes and actual actual show notes and actual transcripts. I think this was the transcripts. I think this was the transcripts. I think this was the transcript one, right? This one is a transcript one, right? This one is a transcript one, right? This one is a good one. Yeah. Okay. So we talked about good one. Yeah. Okay. So we talked about good one. Yeah. Okay. So we talked about all of the things in models. We talked all of the things in models. We talked all of the things in models. We talked about local. Then on agents, we talked about local. Then on agents, we talked about local. Then on agents, we talked about the Azure agent service going. about the Azure agent service going. about the Azure agent service going. talked about the set of oh we forgot to talked about the set of oh we forgot to talked about the set of oh we forgot to mention oh let's look at like all of the mention oh let's look at like all of the mention oh let's look at like all of the things that they can connect to so for things that they can connect to so for things that they can connect to so for example struggling to figure out what example struggling to figure out what example struggling to figure out what did we forget yeah exactly yeah yeah try did we forget yeah exactly yeah yeah try did we forget yeah exactly yeah yeah try it in playground and we can connect them it in playground and we can connect them it in playground and we can connect them to like a whole bunch of different to like a whole bunch of different to like a whole bunch of different things so knowledge Azure I search things so knowledge Azure I search things so knowledge Azure I search Microsoft fabric sharepoint you can do Microsoft fabric sharepoint you can do Microsoft fabric sharepoint you can do grounding with Bing search you can do grounding with Bing search you can do grounding with Bing search you can do custom search for example if you just custom search for example if you just custom search for example if you just want them to go to a specific set of want them to go to a specific set of want them to go to a specific set of sites and not to like entire whole w sites and not to like entire whole w sites and not to like entire whole w then you can just say custom search and then you can just say custom search and then you can just say custom search and then a a few other uh partners and then then a a few other uh partners and then then a a few other uh partners and then it can't be overstated how important it can't be overstated how important it can't be overstated how important grounding is in this kind of context grounding is in this kind of context grounding is in this kind of context because not only excuse me do do I want because not only excuse me do do I want because not only excuse me do do I want to save money and be efficient but I to save money and be efficient but I to save money and be efficient but I don't want my chatbot to talk about don't want my chatbot to talk about don't want my chatbot to talk about things that aren't this things that aren't this things that aren't this it has to be restricted it has to be it has to be restricted it has to be it has to be restricted it has to be grounded and if you're creating a grounded and if you're creating a grounded and if you're creating a chatbot for your own business you're chatbot for your own business you're chatbot for your own business you're going to want to have the same going to want to have the same going to want to have the same experience you don't want your coffee experience you don't want your coffee experience you don't want your coffee shop chatbot offering relationship shop chatbot offering relationship shop chatbot offering relationship advice or opinions about someone's shoes advice or opinions about someone's shoes advice or opinions about someone's shoes it needs to focus on what we're doing it needs to focus on what we're doing it needs to focus on what we're doing here and in this case the Hanselman it's here and in this case the Hanselman it's here and in this case the Hanselman it's one has not only create it into a nano one has not only create it into a nano one has not only create it into a nano model but we're going to ground it in model but we're going to ground it in model but we're going to ground it in the material that we have the knowledge the material that we have the knowledge the material that we have the knowledge that we have available in addition to that we have available in addition to that we have available in addition to knowledge you can add actions so like if knowledge you can add actions so like if knowledge you can add actions so like if you have some deterministic workflows you have some deterministic workflows you have some deterministic workflows that you've already created in Azure that you've already created in Azure that you've already created in Azure logic apps you can bring them in into logic apps you can bring them in into logic apps you can bring them in into connecting to that's a great point so connecting to that's a great point so connecting to that's a great point so this is a thing some people who of a

this is a thing some people who of a this is a thing some people who of a certain age certain age certain age uh might think that well are you uh might think that well are you uh might think that well are you reinventing cron jobs are you reinventing cron jobs are you reinventing cron jobs are you reinventing for loops and things like reinventing for loops and things like reinventing for loops and things like that I have existing processes that work that I have existing processes that work that I have existing processes that work just fine I'm not going to replace them just fine I'm not going to replace them just fine I'm not going to replace them because they work great. But I have because they work great. But I have because they work great. But I have Azure functions, I have Azure Laure Azure functions, I have Azure Laure Azure functions, I have Azure Laure apps. What I'll do is I'll tell the apps. What I'll do is I'll tell the apps. What I'll do is I'll tell the orchestrator agent about those things. orchestrator agent about those things. orchestrator agent about those things. So then it can have those available as So then it can have those available as So then it can have those available as tools. And as long as you can describe tools. And as long as you can describe tools. And as long as you can describe them with an API, you can describe them them with an API, you can describe them them with an API, you can describe them with an open API v3 and then like you with an open API v3 and then like you with an open API v3 and then like you can connect to any of those. So I think can connect to any of those. So I think can connect to any of those. So I think it's time for us to like go back to the it's time for us to like go back to the it's time for us to like go back to the Okay, so let's switch back over to my Okay, so let's switch back over to my Okay, so let's switch back over to my machine. Do you want to call that is machine. Do you want to call that is machine. Do you want to call that is that the the end of the show? I think we that the the end of the show? I think we that the the end of the show? I think we should do that the end of the show. should do that the end of the show. should do that the end of the show. Okay. So, that's the end of this show on Okay. So, that's the end of this show on Okay. So, that's the end of this show on Azure AI Foundry. Uh, hopefully you made Azure AI Foundry. Uh, hopefully you made Azure AI Foundry. Uh, hopefully you made it this far into our podcast. Did you it this far into our podcast. Did you it this far into our podcast. Did you actually say my name correctly on that actually say my name correctly on that actually say my name correctly on that one? Yina with a Y. Arenas. Y I N A. Is one? Yina with a Y. Arenas. Y I N A. Is one? Yina with a Y. Arenas. Y I N A. Is that correctly spelled? That's correct. that correctly spelled? That's correct. that correctly spelled? That's correct. Yes. All right. Cool. This has been Yes. All right. Cool. This has been Yes. All right. Cool. This has been another episode of Hansel Minutes and another episode of Hansel Minutes and another episode of Hansel Minutes and we'll see you again next week. That's we'll see you again next week. That's we'll see you again next week. That's going to be like identified as speaker going to be like identified as speaker going to be like identified as speaker number four.

number four. number four. Okay, we're not done yet. We're just Okay, we're not done yet. We're just Okay, we're not done yet. We're just going to process that through our uh all going to process that through our uh all going to process that through our uh all of the different set of agents that of the different set of agents that of the different set of agents that we've created. Then we're going to go we've created. Then we're going to go we've created. Then we're going to go and talk about our third chapter which and talk about our third chapter which and talk about our third chapter which is all about observability. So I'm going is all about observability. So I'm going is all about observability. So I'm going to kick off a process. You're going to to kick off a process. You're going to to kick off a process. You're going to kick off that going on. I'm copying this kick off that going on. I'm copying this kick off that going on. I'm copying this over putting recording.mpp3. I want to over putting recording.mpp3. I want to over putting recording.mpp3. I want to call out that it was just created and call out that it was just created and call out that it was just created and it's real and then we're going to start it's real and then we're going to start it's real and then we're going to start that kick this process off here. So I'm that kick this process off here. So I'm that kick this process off here. So I'm going to kick off this net process. So going to kick off this net process. So going to kick off this net process. So with this will run for a little while with this will run for a little while with this will run for a little while and I'll let you know when it's done. and I'll let you know when it's done. and I'll let you know when it's done. Okay. Okay. Okay. Okay. So, some of the other things that Okay. So, some of the other things that Okay. So, some of the other things that we saw uh the Azure Oh, we forgot to we saw uh the Azure Oh, we forgot to we saw uh the Azure Oh, we forgot to mention the the Visual Studio Code mention the the Visual Studio Code mention the the Visual Studio Code extension. We can show that. Oh, man. We extension. We can show that. Oh, man. We extension. We can show that. Oh, man. We can show that. Um, so the API, the SDK, can show that. Um, so the API, the SDK, can show that. Um, so the API, the SDK, the Visual Studio Code extension, the Visual Studio Code extension, the Visual Studio Code extension, semantic kernel and autogen more semantic kernel and autogen more semantic kernel and autogen more sessions for you to go and learn about sessions for you to go and learn about sessions for you to go and learn about that. And then we're going to go now to that. And then we're going to go now to that. And then we're going to go now to our final chapter which is around uh our final chapter which is around uh our final chapter which is around uh Foundry Foundry Foundry observability. So for observability, we observability. So for observability, we observability. So for observability, we have the set of tools to support your have the set of tools to support your have the set of tools to support your entire development life cycle. Whether entire development life cycle. Whether entire development life cycle. Whether you are like starting to experiment or you are like starting to experiment or you are like starting to experiment or like thinking about going to production, like thinking about going to production, like thinking about going to production, you want to make sure that you can you want to make sure that you can you want to make sure that you can continuously monitor the quality and continuously monitor the quality and continuously monitor the quality and safety of your solution. So we have safety of your solution. So we have safety of your solution. So we have those tools right here for you to so those tools right here for you to so those tools right here for you to so that you can generate the traces, that you can generate the traces, that you can generate the traces, generate the send it over to Azure generate the send it over to Azure generate the send it over to Azure monitor and like have all of those monitor and like have all of those monitor and like have all of those capabilities that enable enabled you to capabilities that enable enabled you to capabilities that enable enabled you to do debugging and make sure that there's do debugging and make sure that there's do debugging and make sure that there's uh you hit the right level of uh you hit the right level of uh you hit the right level of reliability that you want for your app.

reliability that you want for your app. reliability that you want for your app. Now does this use like hotel open Now does this use like hotel open Now does this use like hotel open telemetry as part of it? Yes. And it's telemetry as part of it? Yes. And it's telemetry as part of it? Yes. And it's important to call using open telemetry important to call using open telemetry important to call using open telemetry hotel is one of the biggest things for hotel is one of the biggest things for hotel is one of the biggest things for folks that are in the observability folks that are in the observability folks that are in the observability space if you're not familiar with open space if you're not familiar with open space if you're not familiar with open telemetry. It is this kind of quiet telemetry. It is this kind of quiet telemetry. It is this kind of quiet storm that is just making everything storm that is just making everything storm that is just making everything better. In the old days, it was just log better. In the old days, it was just log better. In the old days, it was just log files and GP and a regular expression, files and GP and a regular expression, files and GP and a regular expression, right? And that was how we observed our right? And that was how we observed our right? And that was how we observed our software. And everyone knows that if software. And everyone knows that if software. And everyone knows that if you've got a problem and you've decided you've got a problem and you've decided you've got a problem and you've decided to use a regular expression to solve it, to use a regular expression to solve it, to use a regular expression to solve it, now you've got two now you've got two now you've got two problems. What open telemetry does is problems. What open telemetry does is problems. What open telemetry does is it's effectively distributed log files. it's effectively distributed log files. it's effectively distributed log files. And when you have large complicated And when you have large complicated And when you have large complicated distributed systems like this, you want distributed systems like this, you want distributed systems like this, you want all that to come to one place. So you all that to come to one place. So you all that to come to one place. So you could even use a net aspire dashboard or could even use a net aspire dashboard or could even use a net aspire dashboard or graphana or any of these systems any graphana or any of these systems any graphana or any of these systems any existing observability system you have existing observability system you have existing observability system you have you could observe the work of your you could observe the work of your you could observe the work of your agents. So when it comes to agents agents. So when it comes to agents agents. So when it comes to agents agents have threats and run thread is agents have threats and run thread is agents have threats and run thread is the common memory for that conversation the common memory for that conversation the common memory for that conversation and each run is the turns of that the and each run is the turns of that the and each run is the turns of that the agent like a conversational thread. agent like a conversational thread. agent like a conversational thread. Exactly. And every time that the thread Exactly. And every time that the thread Exactly. And every time that the thread is going through, we can see not only is going through, we can see not only is going through, we can see not only the inputs and outputs and the metadata the inputs and outputs and the metadata the inputs and outputs and the metadata that is being generated across every that is being generated across every that is being generated across every single call, but we can also see the single call, but we can also see the single call, but we can also see the evaluations coming in. So for this evaluations coming in. So for this evaluations coming in. So for this particular case, we turned on particular case, we turned on particular case, we turned on evaluations around intent resolution, evaluations around intent resolution, evaluations around intent resolution, relevance, call and code vulnerability.

relevance, call and code vulnerability. relevance, call and code vulnerability. We have new set of uh evaluations as We have new set of uh evaluations as We have new set of uh evaluations as well that are uh releasing as part of well that are uh releasing as part of well that are uh releasing as part of like uh the announcements that we're like uh the announcements that we're like uh the announcements that we're going at at build that include intent going at at build that include intent going at at build that include intent resolution as well but also task resolution as well but also task resolution as well but also task adherence because you want to make sure adherence because you want to make sure adherence because you want to make sure that your agents actually stay on task that your agents actually stay on task that your agents actually stay on task and they're not going around just doing and they're not going around just doing and they're not going around just doing uh other things or for example indirect uh other things or for example indirect uh other things or for example indirect jailbreak where if you have an agent jailbreak where if you have an agent jailbreak where if you have an agent that is going out and getting that is going out and getting that is going out and getting information from the web, what if it information from the web, what if it information from the web, what if it gets some a piece of code that it gets some a piece of code that it gets some a piece of code that it shouldn't. Right. Right. Like so all of shouldn't. Right. Right. Like so all of shouldn't. Right. Right. Like so all of these different set of things we have these different set of things we have these different set of things we have built in. Well, so in the example built in. Well, so in the example built in. Well, so in the example earlier where we went out to the web to earlier where we went out to the web to earlier where we went out to the web to gain information, we spelled your name gain information, we spelled your name gain information, we spelled your name wrong. I spelled your name, sorry, and wrong. I spelled your name, sorry, and wrong. I spelled your name, sorry, and then we got the wrong Yina Gina. Um, at then we got the wrong Yina Gina. Um, at then we got the wrong Yina Gina. Um, at some point we could have a score some point we could have a score some point we could have a score assigned to that and it would say, I'm assigned to that and it would say, I'm assigned to that and it would say, I'm not sure if I'm confident in that. not sure if I'm confident in that. not sure if I'm confident in that. Exactly. It will tell you it will it Exactly. It will tell you it will it Exactly. It will tell you it will it will have a lower quality uh score on, will have a lower quality uh score on, will have a lower quality uh score on, for example, relevance. This one was for example, relevance. This one was for example, relevance. This one was four out of five. It feels that it gave four out of five. It feels that it gave four out of five. It feels that it gave a decent response, but like it could a decent response, but like it could a decent response, but like it could have been better. It does not omit any have been better. It does not omit any have been better. It does not omit any key details and deserves a high score. key details and deserves a high score. key details and deserves a high score. Okay, that's cool. Yeah, one of the Okay, that's cool. Yeah, one of the Okay, that's cool. Yeah, one of the other things that we have in addition to other things that we have in addition to other things that we have in addition to that is integration with your CI/CD that is integration with your CI/CD that is integration with your CI/CD pipelines. So we have uh here we can see pipelines. So we have uh here we can see pipelines. So we have uh here we can see we have the evaluations. So we have the we have the evaluations. So we have the we have the evaluations. So we have the evaluation overview where we've bring in evaluation overview where we've bring in evaluation overview where we've bring in the different set of evals that we've the different set of evals that we've the different set of evals that we've run on our agent and is fully integrated run on our agent and is fully integrated run on our agent and is fully integrated with GitHub as a GitHub action. So every with GitHub as a GitHub action. So every with GitHub as a GitHub action. So every time that you're pushing new um changes time that you're pushing new um changes time that you're pushing new um changes into production, you can run your into production, you can run your into production, you can run your pipelines and like have those those pipelines and like have those those pipelines and like have those those evaluations run for you as well. And evaluations run for you as well. And evaluations run for you as well. And then we have integration with um with

then we have integration with um with then we have integration with um with monitoring as well like where you can monitoring as well like where you can monitoring as well like where you can see all of the operational metrics see all of the operational metrics see all of the operational metrics including your tokens like the average including your tokens like the average including your tokens like the average inference call on your uh duration of inference call on your uh duration of inference call on your uh duration of your inferences goals the errors like your inferences goals the errors like your inferences goals the errors like you we can see how we were in this you we can see how we were in this you we can see how we were in this particular example our quality particular example our quality particular example our quality significantly improved uh over the last significantly improved uh over the last significantly improved uh over the last you know this last few days as we were you know this last few days as we were you know this last few days as we were working hard as we've been working on working hard as we've been working on working hard as we've been working on this demo. Yeah. And uh so like all of this demo. Yeah. And uh so like all of this demo. Yeah. And uh so like all of these is also integrated with Azure these is also integrated with Azure these is also integrated with Azure monitor. So if you have your dashboards, monitor. So if you have your dashboards, monitor. So if you have your dashboards, you can go there and see um all of your you can go there and see um all of your you can go there and see um all of your Yeah, I wanted to call that out. This Yeah, I wanted to call that out. This Yeah, I wanted to call that out. This integrates with Azure monitors integrates with Azure monitors integrates with Azure monitors application insights which is something application insights which is something application insights which is something I'm already familiar with. Now we talked I'm already familiar with. Now we talked I'm already familiar with. Now we talked a little bit about IAS infrastructure as a little bit about IAS infrastructure as a little bit about IAS infrastructure as a service and we talked about platform a service and we talked about platform a service and we talked about platform as a service which is where we are as a service which is where we are as a service which is where we are sitting right now in the AI landscape sitting right now in the AI landscape sitting right now in the AI landscape and then we talked about software as a and then we talked about software as a and then we talked about software as a service where you're going to see things service where you're going to see things service where you're going to see things like copilot. I have found in my time as like copilot. I have found in my time as like copilot. I have found in my time as a cloud person that I really sit in that a cloud person that I really sit in that a cloud person that I really sit in that platform as a service bot. I run my platform as a service bot. I run my platform as a service bot. I run my websites on Azure app service. I use websites on Azure app service. I use websites on Azure app service. I use Azure application um insights. This fits Azure application um insights. This fits Azure application um insights. This fits for me in the kind of stuff that I was for me in the kind of stuff that I was for me in the kind of stuff that I was already doing which is why I was excited already doing which is why I was excited already doing which is why I was excited to be uh a part of this talk.

to be uh a part of this talk. to be uh a part of this talk. Platform as a service is a sweet spot Platform as a service is a sweet spot Platform as a service is a sweet spot that we will mention which is very very that we will mention which is very very that we will mention which is very very important is what do we do around end important is what do we do around end important is what do we do around end toend security right and the integration toend security right and the integration toend security right and the integration with foundry across all of the different with foundry across all of the different with foundry across all of the different stack of Microsoft security products. stack of Microsoft security products. stack of Microsoft security products. First agents that you're creating are First agents that you're creating are First agents that you're creating are getting an specific identity on entra. getting an specific identity on entra. getting an specific identity on entra. So it's not like uh so if you for those So it's not like uh so if you for those So it's not like uh so if you for those of you familiar to ash with ashure of you familiar to ash with ashure of you familiar to ash with ashure active with entra now renamed ashure active with entra now renamed ashure active with entra now renamed ashure active directory with entra um they all active directory with entra um they all active directory with entra um they all of them all of the different set of of them all of the different set of of them all of the different set of entities have whether it is users groups entities have whether it is users groups entities have whether it is users groups devices they have an ident they're devices they have an ident they're devices they have an ident they're registering the directory right now registering the directory right now registering the directory right now agents are going to have their own type agents are going to have their own type agents are going to have their own type of ID where you can assign entitlements of ID where you can assign entitlements of ID where you can assign entitlements and do governance on top of them so you and do governance on top of them so you and do governance on top of them so you can see okay I I'm giving this agent can see okay I I'm giving this agent can see okay I I'm giving this agent who's acting on my behalf a specific set who's acting on my behalf a specific set who's acting on my behalf a specific set of permissions and not more than that of permissions and not more than that of permissions and not more than that the second one is around sending all of the second one is around sending all of the second one is around sending all of the information to purview. So like the information to purview. So like the information to purview. So like things like data labeling, data things like data labeling, data things like data labeling, data protection and the full integration with protection and the full integration with protection and the full integration with all of the security products like all of the security products like all of the security products like Perview and Defender are part of like Perview and Defender are part of like Perview and Defender are part of like what we have uh as as capabilities in what we have uh as as capabilities in what we have uh as as capabilities in Foundry. All righty, we're going to uh Foundry. All righty, we're going to uh Foundry. All righty, we're going to uh these are the sessions around these are the sessions around these are the sessions around observability and governance. Please observability and governance. Please observability and governance. Please take a look. We're going to go way deep take a look. We're going to go way deep take a look. We're going to go way deep into into the sessions. And now let's into into the sessions. And now let's into into the sessions. And now let's don't forget to go back and Okay, let's don't forget to go back and Okay, let's don't forget to go back and Okay, let's switch back over to the factory. One, switch back over to the factory. One, switch back over to the factory. One, two, three. So, I'm a little two, three. So, I'm a little two, three. So, I'm a little sad. Your name is so challenging sad. Your name is so challenging sad. Your name is so challenging apparently. Straight to apparently. Straight to apparently. Straight to Yale. Nobody got that reference, but

Yale. Nobody got that reference, but Yale. Nobody got that reference, but that's fine. Um, so it's when it that's fine. Um, so it's when it that's fine. Um, so it's when it generated that we discussed the generated that we discussed the generated that we discussed the platform's capabilities, models, agents, platform's capabilities, models, agents, platform's capabilities, models, agents, and observability tools. I shared my and observability tools. I shared my and observability tools. I shared my experience streamlining into the experience streamlining into the experience streamlining into the production process and we discussed production process and we discussed production process and we discussed various features. So this is all other various features. So this is all other various features. So this is all other than the misspelling of your name. than the misspelling of your name. than the misspelling of your name. Totally legit and we've got our topics. Totally legit and we've got our topics. Totally legit and we've got our topics. Then it went and it generated a look it Then it went and it generated a look it Then it went and it generated a look it spelled that correct. That's cool. Made spelled that correct. That's cool. Made spelled that correct. That's cool. Made a VTT which is a not just a text file a VTT which is a not just a text file a VTT which is a not just a text file but an actual transcript file that I but an actual transcript file that I but an actual transcript file that I could use on YouTube or I could put into could use on YouTube or I could put into could use on YouTube or I could put into um my podcast. You can see our different um my podcast. You can see our different um my podcast. You can see our different agents over here. agents over here. agents over here. [Laughter] [Laughter] [Laughter] that agent has a a leading space. And that agent has a a leading space. And that agent has a a leading space. And then we have our link verifier which is then we have our link verifier which is then we have our link verifier which is going to check all the links. So we going to check all the links. So we going to check all the links. So we scroll down a little bit. scroll down a little bit. scroll down a little bit. I like a nice asky table. I just feel I like a nice asky table. I just feel I like a nice asky table. I just feel strongly about that. Oh, did did we say strongly about that. Oh, did did we say strongly about that. Oh, did did we say it was special or does it just think it was special or does it just think it was special or does it just think we're special? We are special. We also we're special? We are special. We also we're special? We are special. We also said in this special episode nearly said in this special episode nearly said in this special episode nearly thousand. Okay, so this is all good. So thousand. Okay, so this is all good. So thousand. Okay, so this is all good. So it gave us a summary generating the it gave us a summary generating the it gave us a summary generating the notes. I'm going to open those notes up.

notes. I'm going to open those notes up. notes. I'm going to open those notes up. Then we go and verify each of these. Then we go and verify each of these. Then we go and verify each of these. Looks like two of these didn't work and Looks like two of these didn't work and Looks like two of these didn't work and they have a warning. We'll go and dig they have a warning. We'll go and dig they have a warning. We'll go and dig into those links and then it went and into those links and then it went and into those links and then it went and summarized them. So, let's jump into the summarized them. So, let's jump into the summarized them. So, let's jump into the output output output folder. Here's the show notes and the folder. Here's the show notes and the folder. Here's the show notes and the transcription. We'll go and look at show notes. And then let's do a little notes. And then let's do a little markdown markdown markdown magic. magic. magic. Ooh, scrumptious. Ooh, scrumptious. Ooh, scrumptious. Takeaways. Did we teach it to do Takeaways. Did we teach it to do Takeaways. Did we teach it to do quotes? There we go. Everything we built quotes? There we go. Everything we built quotes? There we go. Everything we built here is legit. It is indeed. It is pretty legit. Less It is indeed. It is pretty legit. Less expensive. That's good. Putting router expensive. That's good. Putting router expensive. That's good. Putting router in the front. Grounding. Yeah, this is in the front. Grounding. Yeah, this is in the front. Grounding. Yeah, this is all cool. Uh, resources mentioned all cool. Uh, resources mentioned all cool. Uh, resources mentioned Scott's YouTube channel. No link. Oh. Scott's YouTube channel. No link. Oh. Scott's YouTube channel. No link. Oh. Ah, scandalous. Oh, and then I don't Ah, scandalous. Oh, and then I don't Ah, scandalous. Oh, and then I don't know if I told you this or not. I know if I told you this or not. I know if I told you this or not. I actually I did not. I I I I was I actually I did not. I I I I was I actually I did not. I I I I was I actually was given permission last night actually was given permission last night actually was given permission last night at 11:30 at night to let people know at 11:30 at night to let people know at 11:30 at night to let people know that Notepad is going to support that Notepad is going to support that Notepad is going to support Markdown. There you go.

So, we've got uh our lovely thing loaded So, we've got uh our lovely thing loaded into just I feel like the AI foundry into just I feel like the AI foundry into just I feel like the AI foundry notepad collaboration. Yeah, I love notepad collaboration. Yeah, I love notepad collaboration. Yeah, I love that. I like that. I love that that. I like that. I love that that. I like that. I love that collaboration. We're going to make collaboration. We're going to make collaboration. We're going to make Markdown. We got to use a little Markdown. We got to use a little Markdown. We got to use a little something that Yeah, that's pretty cool. something that Yeah, that's pretty cool. something that Yeah, that's pretty cool. All right, I dig it. All righty. So, All right, I dig it. All righty. So, All right, I dig it. All righty. So, that got me ourselves a show. Yeah, that that got me ourselves a show. Yeah, that that got me ourselves a show. Yeah, that saved me a lot of work. saved me a lot of work. saved me a lot of work. What are you going to do with that, What are you going to do with that, What are you going to do with that, Scott? I'm going to go and add a couple Scott? I'm going to go and add a couple Scott? I'm going to go and add a couple of small changes because I'm the human of small changes because I'm the human of small changes because I'm the human in the loop and I'm going to give it to in the loop and I'm going to give it to in the loop and I'm going to give it to Mandy who will edit the show and we'll Mandy who will edit the show and we'll Mandy who will edit the show and we'll try to put it up uh as soon as we can. try to put it up uh as soon as we can. try to put it up uh as soon as we can. Sounds good. Very cool. All righty. Sounds good. Very cool. All righty. Sounds good. Very cool. All righty. Yeah. So, let's recap what we did. Let's Yeah. So, let's recap what we did. Let's Yeah. So, let's recap what we did. Let's do a recap and talk about uh where where do a recap and talk about uh where where do a recap and talk about uh where where people can learn more. So remember that people can learn more. So remember that people can learn more. So remember that we tried to create this podcast factory we tried to create this podcast factory we tried to create this podcast factory that would make my life easier. that would make my life easier. that would make my life easier. Hopefully the two or three hours a week Hopefully the two or three hours a week Hopefully the two or three hours a week that I spend on this turns into 10 or 15 that I spend on this turns into 10 or 15 that I spend on this turns into 10 or 15 minutes and just an agent that does this minutes and just an agent that does this minutes and just an agent that does this work for me and then I'll double check work for me and then I'll double check work for me and then I'll double check uh the work. Um one of the things we uh the work. Um one of the things we uh the work. Um one of the things we didn't get a chance to show you were didn't get a chance to show you were didn't get a chance to show you were some ideas on guest sourcing because I some ideas on guest sourcing because I some ideas on guest sourcing because I realized that there's a trend line on my realized that there's a trend line on my realized that there's a trend line on my show and we can get recommendations show and we can get recommendations show and we can get recommendations about who else I might want to have on about who else I might want to have on about who else I might want to have on the show. shocking that it hadn't the show. shocking that it hadn't the show. shocking that it hadn't recommended you until just now. You may recommended you until just now. You may recommended you until just now. You may have uh put your thumb on the scale in have uh put your thumb on the scale in have uh put your thumb on the scale in the AI. We got the biogenerator and then the AI. We got the biogenerator and then the AI. We got the biogenerator and then we also could have an AI doing uh we also could have an AI doing uh we also could have an AI doing uh scheduling because it's really scheduling because it's really scheduling because it's really challenging. There's a lot of emails challenging. There's a lot of emails challenging. There's a lot of emails going back and forth talking to to going back and forth talking to to going back and forth talking to to guests trying to get them scheduled for guests trying to get them scheduled for guests trying to get them scheduled for the show. We've got a transcript the show. We've got a transcript the show. We've got a transcript generator that we saw which is a custom generator that we saw which is a custom generator that we saw which is a custom model and show notes which was done on model and show notes which was done on model and show notes which was done on GPT41 nano show notes the fine-tune GPT41 nano show notes the fine-tune GPT41 nano show notes the fine-tune model fine-tune model and then the link model fine-tune model and then the link model fine-tune model and then the link resolver make sure that the links that resolver make sure that the links that resolver make sure that the links that it puts up are legit links make sure

it puts up are legit links make sure it puts up are legit links make sure that they don't 404 and then other that they don't 404 and then other that they don't 404 and then other things that we could potentially do things that we could potentially do things that we could potentially do would be generate copy for social and would be generate copy for social and would be generate copy for social and things like that localization has been a things like that localization has been a things like that localization has been a really interesting thing we've got really interesting thing we've got really interesting thing we've got models that could do that as well so I models that could do that as well so I models that could do that as well so I could translation bringing to different could translation bringing to different could translation bringing to different languages that's another thing that we languages that's another thing that we languages that's another thing that we could have Mhm. All righty. Well, with could have Mhm. All righty. Well, with could have Mhm. All righty. Well, with that, this is a way in which we you that, this is a way in which we you that, this is a way in which we you hopefully saw today how Azure II foundry hopefully saw today how Azure II foundry hopefully saw today how Azure II foundry can help infuse AI into like a different can help infuse AI into like a different can help infuse AI into like a different set of applications. We build a podcast set of applications. We build a podcast set of applications. We build a podcast factory. What factory are you going to factory. What factory are you going to factory. What factory are you going to build? How are you going to like bring build? How are you going to like bring build? How are you going to like bring that uh investment to be instead of that uh investment to be instead of that uh investment to be instead of return of investment return on your return of investment return on your return of investment return on your effort? We're hoping that like you know effort? We're hoping that like you know effort? We're hoping that like you know you're going to see how AI can help you're going to see how AI can help you're going to see how AI can help reduce toil, can help re like take away reduce toil, can help re like take away reduce toil, can help re like take away all of these set of tasks that we don't all of these set of tasks that we don't all of these set of tasks that we don't necessarily want to do. You mentioned necessarily want to do. You mentioned necessarily want to do. You mentioned dump, dangerous and uh dull, dangerous dump, dangerous and uh dull, dangerous dump, dangerous and uh dull, dangerous and dull. I want to do the fun stuff. and dull. I want to do the fun stuff. and dull. I want to do the fun stuff. And then we have a lot of customers like And then we have a lot of customers like And then we have a lot of customers like Gainsite, Safeer, Henkin, and more that Gainsite, Safeer, Henkin, and more that Gainsite, Safeer, Henkin, and more that are using AI today. Over 70,000 are using AI today. Over 70,000 are using AI today. Over 70,000 customers are using Azure AI foundry to customers are using Azure AI foundry to customers are using Azure AI foundry to bring AI into their app into their bring AI into their app into their bring AI into their app into their organizations. over 10,000 have already organizations. over 10,000 have already organizations. over 10,000 have already used the Azure Foundry agent service and used the Azure Foundry agent service and used the Azure Foundry agent service and are creating now millions and millions are creating now millions and millions are creating now millions and millions of agents to like bring in and get of agents to like bring in and get of agents to like bring in and get significant input uh significant gains significant input uh significant gains significant input uh significant gains into their into their into their processes. We want to leave you with a processes. We want to leave you with a processes. We want to leave you with a call to call to call to action. Are you ready to create the action. Are you ready to create the action. Are you ready to create the future of future of future of AI? Thank you folks. Yeah. Thank you

AI? Thank you folks. Yeah. Thank you AI? Thank you folks. Yeah. Thank you very very very much. Take a picture. much. Take a picture. much. Take a picture. take a picture of that and make sure take a picture of that and make sure take a picture of that and make sure that you remember all of the notable that you remember all of the notable that you remember all of the notable quotes from uh from the show and then uh quotes from uh from the show and then uh quotes from uh from the show and then uh sign up over here my sign up over here my sign up over here my friends. Thank you. Thank you folks. friends. Thank you. Thank you folks. friends. Thank you. Thank you folks. [Music]