APIs as the Execution Layer of AI with Postman's Rodric Rabbah

Hi friends, I'm Scott Hansselman. It's Hi friends, I'm Scott Hansselman. It's another episode of Hansel Minutes. Today another episode of Hansel Minutes. Today another episode of Hansel Minutes. Today I'm talking with Rodrik Rabba. He's the I'm talking with Rodrik Rabba. He's the I'm talking with Rodrik Rabba. He's the head of product of Postman Flows. How head of product of Postman Flows. How head of product of Postman Flows. How are you sir? are you sir? are you sir? >> I'm good Scott. Thanks for having me on. >> I'm good Scott. Thanks for having me on. >> I'm good Scott. Thanks for having me on. Really excited for this. Really excited for this. Really excited for this. >> Yeah, thanks for hanging hang hanging >> Yeah, thanks for hanging hang hanging >> Yeah, thanks for hanging hang hanging out with me. you know, you have done a out with me. you know, you have done a out with me. you know, you have done a bunch of work in the cloud since the bunch of work in the cloud since the bunch of work in the cloud since the cloud's been around. Uh, I'm curious cloud's been around. Uh, I'm curious cloud's been around. Uh, I'm curious when you were working as a research when you were working as a research when you were working as a research scientist. Did you see the cloud coming? scientist. Did you see the cloud coming? scientist. Did you see the cloud coming? Did you see serverless coming or did it Did you see serverless coming or did it Did you see serverless coming or did it happen and then you jumped on board? happen and then you jumped on board? happen and then you jumped on board? >> Uh, it was sort of happening and you >> Uh, it was sort of happening and you >> Uh, it was sort of happening and you realized this was a big thing and jumped realized this was a big thing and jumped realized this was a big thing and jumped on board. Um, definitely did not uh I on board. Um, definitely did not uh I on board. Um, definitely did not uh I moved into cloud. I was working on moved into cloud. I was working on moved into cloud. I was working on compilers, runtime systems, building compilers, runtime systems, building compilers, runtime systems, building software for FPGAs, basically software for FPGAs, basically software for FPGAs, basically programmable hardware. Um, and my foray programmable hardware. Um, and my foray programmable hardware. Um, and my foray into cloud started with Amazon Lambda. into cloud started with Amazon Lambda. into cloud started with Amazon Lambda. When it came on, it basically um it gave When it came on, it basically um it gave When it came on, it basically um it gave us the vocabulary for uh how to present us the vocabulary for uh how to present us the vocabulary for uh how to present and tell the story around what we were and tell the story around what we were and tell the story around what we were doing uh around high level synthesis for doing uh around high level synthesis for doing uh around high level synthesis for hardware and uh started working in cloud hardware and uh started working in cloud hardware and uh started working in cloud and off I went and never looked back in and off I went and never looked back in and off I went and never looked back in a way. I think I've been on a mission to a way. I think I've been on a mission to a way. I think I've been on a mission to build a cloud for Rodri ever since.

build a cloud for Rodri ever since. build a cloud for Rodri ever since. >> Yeah, build the Rodri cloud. >> Yeah, build the Rodri cloud. >> Yeah, build the Rodri cloud. It feels like, and this is an this is a It feels like, and this is an this is a It feels like, and this is an this is a um an analysis of your career in the um an analysis of your career in the um an analysis of your career in the research that I've done that you've been research that I've done that you've been research that I've done that you've been looking for the right layer of looking for the right layer of looking for the right layer of abstraction for people to work at. I abstraction for people to work at. I abstraction for people to work at. I mean, you pioneered serverless computing mean, you pioneered serverless computing mean, you pioneered serverless computing when you worked on Open Whisk at IBM. when you worked on Open Whisk at IBM. when you worked on Open Whisk at IBM. You founded a serverless cloud company. You founded a serverless cloud company. You founded a serverless cloud company. Serverless is where you think about Serverless is where you think about Serverless is where you think about servers less uh and now you're working servers less uh and now you're working servers less uh and now you're working on this AI to API paradigm shift. Uh it on this AI to API paradigm shift. Uh it on this AI to API paradigm shift. Uh it seems like you're hunting for the right seems like you're hunting for the right seems like you're hunting for the right layer of abstraction for the business layer of abstraction for the business layer of abstraction for the business user to get work done, for the developer user to get work done, for the developer user to get work done, for the developer to get work done. Is that a fair to get work done. Is that a fair to get work done. Is that a fair analysis? analysis? analysis? >> I I think that's a that's a really good >> I I think that's a that's a really good >> I I think that's a that's a really good analysis and good way of looking at analysis and good way of looking at analysis and good way of looking at essentially my own personal mission. Um essentially my own personal mission. Um essentially my own personal mission. Um I still build I build all the time and I still build I build all the time and I still build I build all the time and in a way that puts me in touch with the in a way that puts me in touch with the in a way that puts me in touch with the technology that um not only our users on technology that um not only our users on technology that um not only our users on Postman Postman flows use but in general Postman Postman flows use but in general Postman Postman flows use but in general what engineers are dealing with and the what engineers are dealing with and the what engineers are dealing with and the power and intersection of APIs and AI power and intersection of APIs and AI power and intersection of APIs and AI today is really transformative. Um like today is really transformative. Um like today is really transformative. Um like I've seen it firsthand. Um, I think my I've seen it firsthand. Um, I think my I've seen it firsthand. Um, I think my first interaction between APIs and AI first interaction between APIs and AI first interaction between APIs and AI was like when I first connected an API was like when I first connected an API was like when I first connected an API to a large language model, Chachi to a large language model, Chachi to a large language model, Chachi Clawude actually. Uh, I was like, "Oh my Clawude actually. Uh, I was like, "Oh my Clawude actually. Uh, I was like, "Oh my god, I was like, more APIs, more APIs. I god, I was like, more APIs, more APIs. I god, I was like, more APIs, more APIs. I need to just give it more APIs." Uh, but need to just give it more APIs." Uh, but need to just give it more APIs." Uh, but I think hunting for the right I think hunting for the right I think hunting for the right abstraction um is a really good way of abstraction um is a really good way of abstraction um is a really good way of putting it. you know you started you putting it. you know you started you putting it. you know you started you know I said I started in hardware and

know I said I started in hardware and know I said I started in hardware and basically just building compilers which basically just building compilers which basically just building compilers which is the lowest level of abstraction and is the lowest level of abstraction and is the lowest level of abstraction and increasingly you've been going up the increasingly you've been going up the increasingly you've been going up the level uh the stack uh because at the end level uh the stack uh because at the end level uh the stack uh because at the end of the day as builders engineers today's of the day as builders engineers today's of the day as builders engineers today's definition of a builder is very broad definition of a builder is very broad definition of a builder is very broad anybody who has an idea and access to um anybody who has an idea and access to um anybody who has an idea and access to um you know claw ch etc could build um so you know claw ch etc could build um so you know claw ch etc could build um so how do you make them efficient how do how do you make them efficient how do how do you make them efficient how do you make them uh capable of realizing you make them uh capable of realizing you make them uh capable of realizing what's in their head and actually what's in their head and actually what's in their head and actually materializing it. Um, and increasingly materializing it. Um, and increasingly materializing it. Um, and increasingly it's sort of the power of APIs and AI it's sort of the power of APIs and AI it's sort of the power of APIs and AI really brings it to a four for me. really brings it to a four for me. really brings it to a four for me. >> Yeah. I I think that people right now >> Yeah. I I think that people right now >> Yeah. I I think that people right now are hearing the word agents. If you want are hearing the word agents. If you want are hearing the word agents. If you want to feel smart in a meeting, you say to feel smart in a meeting, you say to feel smart in a meeting, you say agentic multiple times. It's a very kind agentic multiple times. It's a very kind agentic multiple times. It's a very kind of aerodite, very kind of academic word. of aerodite, very kind of academic word. of aerodite, very kind of academic word. People love to say agentic and then feel People love to say agentic and then feel People love to say agentic and then feel smart. Um, but ultimately tool calling smart. Um, but ultimately tool calling smart. Um, but ultimately tool calling is what we're talking about. And for is what we're talking about. And for is what we're talking about. And for APIs, we're tool calling into some APIs, we're tool calling into some APIs, we're tool calling into some amazing backend. It feels like Postman amazing backend. It feels like Postman amazing backend. It feels like Postman really did a brilliant kind of not even really did a brilliant kind of not even really did a brilliant kind of not even a pivot. They just kind of like stepped a pivot. They just kind of like stepped a pivot. They just kind of like stepped in and said, "Hey, we're the API in and said, "Hey, we're the API in and said, "Hey, we're the API company." I thought that was very clever company." I thought that was very clever company." I thought that was very clever because I've been using REST. I I did because I've been using REST. I I did because I've been using REST. I I did SOAP. I did Cororba. I did decom. Like SOAP. I did Cororba. I did decom. Like SOAP. I did Cororba. I did decom. Like we've always been trying to figure out we've always been trying to figure out we've always been trying to figure out what the right level of distributed API what the right level of distributed API what the right level of distributed API is. It seems like Postman flows is an is. It seems like Postman flows is an is. It seems like Postman flows is an acknowledgment that hey, we have all acknowledgment that hey, we have all acknowledgment that hey, we have all these APIs. Those are just tool calls.

these APIs. Those are just tool calls. these APIs. Those are just tool calls. We could plug the whole world into We could plug the whole world into We could plug the whole world into agents and into LLMs and look what kind agents and into LLMs and look what kind agents and into LLMs and look what kind of cool stuff we can do. Oh, absolutely. of cool stuff we can do. Oh, absolutely. of cool stuff we can do. Oh, absolutely. I think the the irony here is we're I think the the irony here is we're I think the the irony here is we're inventing new terms uh which maybe inventing new terms uh which maybe inventing new terms uh which maybe confuses people uh or maybe simplifies confuses people uh or maybe simplifies confuses people uh or maybe simplifies uh for new builders. Like you say API, uh for new builders. Like you say API, uh for new builders. Like you say API, maybe it might scare somebody off who's maybe it might scare somebody off who's maybe it might scare somebody off who's not a typical developer. You say tools not a typical developer. You say tools not a typical developer. You say tools like okay tool helps me get the job like okay tool helps me get the job like okay tool helps me get the job done. Um and I think what's intriguing done. Um and I think what's intriguing done. Um and I think what's intriguing with with AI today and largely we're with with AI today and largely we're with with AI today and largely we're talking largely about large language talking largely about large language talking largely about large language models uh here which have a profound models uh here which have a profound models uh here which have a profound capability of understanding our written capability of understanding our written capability of understanding our written language um is that you could you give language um is that you could you give language um is that you could you give it these tools essentially accomplish it these tools essentially accomplish it these tools essentially accomplish tasks for you otherwise you're just tasks for you otherwise you're just tasks for you otherwise you're just having a text in text out conversation having a text in text out conversation having a text in text out conversation uh with Chad GPT or Claude or whatever uh with Chad GPT or Claude or whatever uh with Chad GPT or Claude or whatever is your Gemini uh another tool that I is your Gemini uh another tool that I is your Gemini uh another tool that I use actually got my kids using use actually got my kids using use actually got my kids using Um, I vibe coded a game the other day. Um, I vibe coded a game the other day. Um, I vibe coded a game the other day. It was very interesting to just see my It was very interesting to just see my It was very interesting to just see my my 11-year-old son, my 14-year-old son my 11-year-old son, my 14-year-old son my 11-year-old son, my 14-year-old son both created different games. Like, both created different games. Like, both created different games. Like, wait, you had to write all this code by wait, you had to write all this code by wait, you had to write all this code by hand and it would have taken you how hand and it would have taken you how hand and it would have taken you how long? And we just did it in 10 minutes.

long? And we just did it in 10 minutes. long? And we just did it in 10 minutes. Uh, but anyway, getting back to the Uh, but anyway, getting back to the Uh, but anyway, getting back to the point, um, you know, when you're point, um, you know, when you're point, um, you know, when you're building these applications, you realize building these applications, you realize building these applications, you realize I need a way of translating my idea, I need a way of translating my idea, I need a way of translating my idea, which I do with natural language. The which I do with natural language. The which I do with natural language. The LLM can understand it. and then to LLM can understand it. and then to LLM can understand it. and then to actually affect change in the world um actually affect change in the world um actually affect change in the world um to send a message to take a payment uh to send a message to take a payment uh to send a message to take a payment uh to render something on a screen all of to render something on a screen all of to render something on a screen all of that happens with APIs or tools in that happens with APIs or tools in that happens with APIs or tools in common parlance um and so for Postman common parlance um and so for Postman common parlance um and so for Postman you know you're calling it clever it was you know you're calling it clever it was you know you're calling it clever it was kind of like we're already here uh like kind of like we're already here uh like kind of like we're already here uh like this is where all the world's APIs are this is where all the world's APIs are this is where all the world's APIs are over 100,000 APIs are on Postman today over 100,000 APIs are on Postman today over 100,000 APIs are on Postman today in terms of our Postman API network in terms of our Postman API network in terms of our Postman API network every one of those APIs is a tool you every one of those APIs is a tool you every one of those APIs is a tool you can attach to your AI I and I think when can attach to your AI I and I think when can attach to your AI I and I think when you start building it's like you give it you start building it's like you give it you start building it's like you give it one API one tool it's like great oh my one API one tool it's like great oh my one API one tool it's like great oh my god it can do so much more tools give it god it can do so much more tools give it god it can do so much more tools give it more tools and hence more APIs and I more tools and hence more APIs and I more tools and hence more APIs and I think that's been the transformation not think that's been the transformation not think that's been the transformation not just for me personally when I started just for me personally when I started just for me personally when I started building with this but in general why building with this but in general why building with this but in general why there's so much excitement in the space there's so much excitement in the space there's so much excitement in the space with artificial intelligence today um with artificial intelligence today um with artificial intelligence today um tools why MCP went became so uh so tools why MCP went became so uh so tools why MCP went became so uh so popular is because just made it easier popular is because just made it easier popular is because just made it easier for you to connect your APIs um to your for you to connect your APIs um to your for you to connect your APIs um to your LLM and just unlock massive capabilities LLM and just unlock massive capabilities LLM and just unlock massive capabilities >> to to ask a somewhat challenging >> to to ask a somewhat challenging >> to to ask a somewhat challenging question and kind of like bring us down question and kind of like bring us down question and kind of like bring us down from like I I get where the re where the from like I I get where the re where the from like I I get where the re where the future is. I get where the goal is, but future is. I get where the goal is, but future is. I get where the goal is, but I have also found LLMs attached to tools I have also found LLMs attached to tools I have also found LLMs attached to tools attached to MCP servers because no one's attached to MCP servers because no one's attached to MCP servers because no one's really figured out how to write a great really figured out how to write a great really figured out how to write a great MCP server. One might ship one and it's

MCP server. One might ship one and it's MCP server. One might ship one and it's got two tools. Another one might ship an got two tools. Another one might ship an got two tools. Another one might ship an MCP server and it's got 85 tools and now MCP server and it's got 85 tools and now MCP server and it's got 85 tools and now it's used up all my tool available it's used up all my tool available it's used up all my tool available tooling space. It feels like someone tooling space. It feels like someone tooling space. It feels like someone taking a uh a screwdriver and and going taking a uh a screwdriver and and going taking a uh a screwdriver and and going shoving like a Phillips screwdriver into shoving like a Phillips screwdriver into shoving like a Phillips screwdriver into a square head and then shoving a square a square head and then shoving a square a square head and then shoving a square head into a Phillips. Um if someone were head into a Phillips. Um if someone were head into a Phillips. Um if someone were to plug an LLM into a Postman API, to plug an LLM into a Postman API, to plug an LLM into a Postman API, some of those are destructive APIs. Like some of those are destructive APIs. Like some of those are destructive APIs. Like fumbling around in the dark poking fumbling around in the dark poking fumbling around in the dark poking things with screwdrivers might work with things with screwdrivers might work with things with screwdrivers might work with a local LLM and a local series of tools, a local LLM and a local series of tools, a local LLM and a local series of tools, but one could charge a credit card or but one could charge a credit card or but one could charge a credit card or make a mistake. make a mistake. make a mistake. >> Absolutely. Doesn't the altitude of our >> Absolutely. Doesn't the altitude of our >> Absolutely. Doesn't the altitude of our APIs and how we think about their APIs and how we think about their APIs and how we think about their behaviors make a even bigger difference behaviors make a even bigger difference behaviors make a even bigger difference than before if the LLM could be fumbling than before if the LLM could be fumbling than before if the LLM could be fumbling around in the dark with the tool? around in the dark with the tool? around in the dark with the tool? >> Oh, absolutely. I I love the fumbling in >> Oh, absolutely. I I love the fumbling in >> Oh, absolutely. I I love the fumbling in the dark analogy. Like everybody has a the dark analogy. Like everybody has a the dark analogy. Like everybody has a toolbox where there's like thousand toolbox where there's like thousand toolbox where there's like thousand tools in or maybe dozen tools in there tools in or maybe dozen tools in there tools in or maybe dozen tools in there and uh you know it's just where you just and uh you know it's just where you just and uh you know it's just where you just keep throwing stuff and so you have a keep throwing stuff and so you have a keep throwing stuff and so you have a job to do, you go reach for it, maybe job to do, you go reach for it, maybe job to do, you go reach for it, maybe you pull out the wrong tool and your job you pull out the wrong tool and your job you pull out the wrong tool and your job becomes much harder to do. The right becomes much harder to do. The right becomes much harder to do. The right tool for the right job makes all the tool for the right job makes all the tool for the right job makes all the difference. That's why we invent uh difference. That's why we invent uh difference. That's why we invent uh these tools. And it's the same with these tools. And it's the same with these tools. And it's the same with LLMs. You give them different tools that LLMs. You give them different tools that LLMs. You give them different tools that are not well structured, uh not are not well structured, uh not are not well structured, uh not described well, where the LLM does not described well, where the LLM does not described well, where the LLM does not know when to use a particular tool, it's know when to use a particular tool, it's know when to use a particular tool, it's going to guess and you're going to going to guess and you're going to going to guess and you're going to suffer the consequences. Um so that's suffer the consequences. Um so that's suffer the consequences. Um so that's why it's important that when you're why it's important that when you're why it's important that when you're attaching tools to your LLM, you give attaching tools to your LLM, you give attaching tools to your LLM, you give them good descriptions. You tell it what them good descriptions. You tell it what them good descriptions. You tell it what parameters it needs to pass in to call parameters it needs to pass in to call parameters it needs to pass in to call your tool. Um and uh your tool had

your tool. Um and uh your tool had your tool. Um and uh your tool had better be well structured. You have you better be well structured. You have you better be well structured. You have you need good safeguards around it. And if need good safeguards around it. And if need good safeguards around it. And if you're a software developer hearing this you're a software developer hearing this you're a software developer hearing this would be like wait this is what I do would be like wait this is what I do would be like wait this is what I do when I build software. Uh absolutely. Um when I build software. Uh absolutely. Um when I build software. Uh absolutely. Um and in some ways you know we tell our and in some ways you know we tell our and in some ways you know we tell our engineers please document your code. Um engineers please document your code. Um engineers please document your code. Um and you do that because for somebody and you do that because for somebody and you do that because for somebody else who comes and looks at your code else who comes and looks at your code else who comes and looks at your code they should be able to understand it you they should be able to understand it you they should be able to understand it you know but we don't maybe do a very good know but we don't maybe do a very good know but we don't maybe do a very good job documenting our code as developers. job documenting our code as developers. job documenting our code as developers. Sometimes we tend to be lazy. I'm Sometimes we tend to be lazy. I'm Sometimes we tend to be lazy. I'm speaking for myself. I tend to be very speaking for myself. I tend to be very speaking for myself. I tend to be very lazy. to my documentation isn't always lazy. to my documentation isn't always lazy. to my documentation isn't always very good. But when you realize that very good. But when you realize that very good. But when you realize that your consumer is an AI and it's very your consumer is an AI and it's very your consumer is an AI and it's very unforgiving sometimes because it's going unforgiving sometimes because it's going unforgiving sometimes because it's going to grab one tool, it's going to do a to grab one tool, it's going to do a to grab one tool, it's going to do a job. It's going to grab another tool, do job. It's going to grab another tool, do job. It's going to grab another tool, do maybe try to do the same job with it. Uh maybe try to do the same job with it. Uh maybe try to do the same job with it. Uh and like you said, the consequences and like you said, the consequences and like you said, the consequences could be dire. The APIs could be could be dire. The APIs could be could be dire. The APIs could be destructive, tear down my database, take destructive, tear down my database, take destructive, tear down my database, take a payment, um send a message that I a payment, um send a message that I a payment, um send a message that I shouldn't have sent to a customer or uh shouldn't have sent to a customer or uh shouldn't have sent to a customer or uh um so we're essentially programmed by AI um so we're essentially programmed by AI um so we're essentially programmed by AI already. like we're changing our already. like we're changing our already. like we're changing our behavior largely because we see the behavior largely because we see the behavior largely because we see the immense power the immense potential that immense power the immense potential that immense power the immense potential that AI and LLM unlock for us. Uh so it's AI and LLM unlock for us. Uh so it's AI and LLM unlock for us. Uh so it's been you know we've been on a decade been you know we've been on a decade been you know we've been on a decade long journey to get engineers and long journey to get engineers and long journey to get engineers and software developers to do better with software developers to do better with software developers to do better with the software that they write or the APIs the software that they write or the APIs the software that they write or the APIs that they create because those APIs that they create because those APIs that they create because those APIs become tools like you know we're become tools like you know we're become tools like you know we're welcoming in our AI overlords. were welcoming in our AI overlords. were welcoming in our AI overlords. were saying, "Okay, we're going to do it in saying, "Okay, we're going to do it in saying, "Okay, we're going to do it in service of the AI because the value that service of the AI because the value that service of the AI because the value that you get out of it is massive." Like the you get out of it is massive." Like the you get out of it is massive." Like the potential we're unlocking with that is potential we're unlocking with that is potential we're unlocking with that is massive. But you have to be like, you massive. But you have to be like, you massive. But you have to be like, you don't throw out the playbook of what you don't throw out the playbook of what you don't throw out the playbook of what you do to build good software. Um, good do to build good software. Um, good do to build good software. Um, good structured code, well- tested code, good

structured code, well- tested code, good structured code, well- tested code, good guardrails, good authentication, good guardrails, good authentication, good guardrails, good authentication, good audit trails, all of those become even audit trails, all of those become even audit trails, all of those become even more important now because you've got more important now because you've got more important now because you've got this AI, you mentioned the word agent this AI, you mentioned the word agent this AI, you mentioned the word agent before, that's essentially autonomous before, that's essentially autonomous before, that's essentially autonomous code that's just going to go and work. code that's just going to go and work. code that's just going to go and work. And sometimes the best analogy is like And sometimes the best analogy is like And sometimes the best analogy is like imagine you have, you know, a child with imagine you have, you know, a child with imagine you have, you know, a child with a bunch of very powerful tools, your a bunch of very powerful tools, your a bunch of very powerful tools, your APIs. Would you trust them to just go APIs. Would you trust them to just go APIs. Would you trust them to just go and grab tools willy-nilly? No, you and grab tools willy-nilly? No, you and grab tools willy-nilly? No, you wouldn't. And so that coaching that you wouldn't. And so that coaching that you wouldn't. And so that coaching that you would do for another person is the same would do for another person is the same would do for another person is the same thing you would have to do for an LLM. thing you would have to do for an LLM. thing you would have to do for an LLM. That's why it's also very relatable to That's why it's also very relatable to That's why it's also very relatable to me. I was like, you can explain this me. I was like, you can explain this me. I was like, you can explain this even to my kids like I was explaining even to my kids like I was explaining even to my kids like I was explaining before. like this is why you do it one before. like this is why you do it one before. like this is why you do it one way and not another. When it doesn't get way and not another. When it doesn't get way and not another. When it doesn't get the outcome that you want, they'll get the outcome that you want, they'll get the outcome that you want, they'll get very frustrated like well you have to very frustrated like well you have to very frustrated like well you have to explain it a little bit better. Uh don't explain it a little bit better. Uh don't explain it a little bit better. Uh don't be lazy in how you explain and ask what be lazy in how you explain and ask what be lazy in how you explain and ask what you want out of it. It will respond but you want out of it. It will respond but you want out of it. It will respond but better input gets you better output. Uh better input gets you better output. Uh better input gets you better output. Uh so it's been very interesting to sort of so it's been very interesting to sort of so it's been very interesting to sort of see that interaction. Uh and I think a see that interaction. Uh and I think a see that interaction. Uh and I think a lot of us are just exploring. We're lot of us are just exploring. We're lot of us are just exploring. We're discovering right now.

discovering right now. discovering right now. >> Yeah. Providing so much more context. I >> Yeah. Providing so much more context. I >> Yeah. Providing so much more context. I I call it yapping. Like I've seen people I call it yapping. Like I've seen people I call it yapping. Like I've seen people go and try to oneshot their job. Hey, go and try to oneshot their job. Hey, go and try to oneshot their job. Hey, make me a marketing deck for blah blah make me a marketing deck for blah blah make me a marketing deck for blah blah blah. That is not a prompt, blah. That is not a prompt, blah. That is not a prompt, >> right? >> right? >> right? >> But sit there and talk to it for 20 >> But sit there and talk to it for 20 >> But sit there and talk to it for 20 minutes and give it several paragraphs minutes and give it several paragraphs minutes and give it several paragraphs and tell it all about yourself and you and tell it all about yourself and you and tell it all about yourself and you might a lot closer to what your ultimate might a lot closer to what your ultimate might a lot closer to what your ultimate goal is, goal is, goal is, >> right? Yeah. I mean, is that vibe >> right? Yeah. I mean, is that vibe >> right? Yeah. I mean, is that vibe coding? I mean, I haven't seen a good coding? I mean, I haven't seen a good coding? I mean, I haven't seen a good definition of vibe coding. Vibe coding I definition of vibe coding. Vibe coding I definition of vibe coding. Vibe coding I I prefer and I think that there's folks I prefer and I think that there's folks I prefer and I think that there's folks here at my company in the day in my day here at my company in the day in my day here at my company in the day in my day job at Microsoft that are calling it job at Microsoft that are calling it job at Microsoft that are calling it specificationdriven development because specificationdriven development because specificationdriven development because ultimately I think that there are some ultimately I think that there are some ultimately I think that there are some really really interesting philosophical really really interesting philosophical really really interesting philosophical conversations that people aren't having conversations that people aren't having conversations that people aren't having where it's like you need to be really where it's like you need to be really where it's like you need to be really really specific. Well, you turn up the really specific. Well, you turn up the really specific. Well, you turn up the specificity on English pros, it turns specificity on English pros, it turns specificity on English pros, it turns into a programming language that you can into a programming language that you can into a programming language that you can compile. So if I if I wanted a if I compile. So if I if I wanted a if I compile. So if I if I wanted a if I wanted an abstract syntax tree of my wanted an abstract syntax tree of my wanted an abstract syntax tree of my pros, I would just write it in C# or pros, I would just write it in C# or pros, I would just write it in C# or Python. Python. Python. >> Right. Right. And and I think that >> Right. Right. And and I think that >> Right. Right. And and I think that that's what's uh I maybe irony is the is that's what's uh I maybe irony is the is that's what's uh I maybe irony is the is the right right word here is like we're the right right word here is like we're the right right word here is like we're saying you're not going to program saying you're not going to program saying you're not going to program anymore. Like yeah, you are you're just anymore. Like yeah, you are you're just anymore. Like yeah, you are you're just programming with a different language programming with a different language programming with a different language and the level of verbosity specificity and the level of verbosity specificity and the level of verbosity specificity that you need to communicate the context that you need to communicate the context that you need to communicate the context is so powerful. But we invented is so powerful. But we invented is so powerful. But we invented programming languages because the programming languages because the programming languages because the abstraction that digital computers give abstraction that digital computers give abstraction that digital computers give us like the zeros and ones is so unlike us like the zeros and ones is so unlike us like the zeros and ones is so unlike anything else in nature. Um and LLMs in anything else in nature. Um and LLMs in anything else in nature. Um and LLMs in a way are just like bring us back to a way are just like bring us back to a way are just like bring us back to what's real. It's like I'm going to yolo

what's real. It's like I'm going to yolo what's real. It's like I'm going to yolo it. It's like I'm going to figure it it. It's like I'm going to figure it it. It's like I'm going to figure it out. And so to guide it and restrict it out. And so to guide it and restrict it out. And so to guide it and restrict it and constrain it, you start essentially and constrain it, you start essentially and constrain it, you start essentially speaking to it in a particular way. And speaking to it in a particular way. And speaking to it in a particular way. And at what point does it become a at what point does it become a at what point does it become a programming language? programming language? programming language? >> Yeah. You know, there was a time a >> Yeah. You know, there was a time a >> Yeah. You know, there was a time a couple years ago where everyone was couple years ago where everyone was couple years ago where everyone was calling themselves a full stack calling themselves a full stack calling themselves a full stack developer and you know, full stack for developer and you know, full stack for developer and you know, full stack for them might have stopped at JavaScript, them might have stopped at JavaScript, them might have stopped at JavaScript, right? And it didn't consider process right? And it didn't consider process right? And it didn't consider process boundaries and threads and all that kind boundaries and threads and all that kind boundaries and threads and all that kind of stuff. And the next guest on this of stuff. And the next guest on this of stuff. And the next guest on this show next week will be Randall Hyde who show next week will be Randall Hyde who show next week will be Randall Hyde who wrote one of the very first 6502 wrote one of the very first 6502 wrote one of the very first 6502 assembly language compilers like in the assembly language compilers like in the assembly language compilers like in the in the middle 70s. And the difference in the middle 70s. And the difference in the middle 70s. And the difference between where the business person wants between where the business person wants between where the business person wants to do their visual low code workflow to do their visual low code workflow to do their visual low code workflow builder and where Randall Hyde's brain builder and where Randall Hyde's brain builder and where Randall Hyde's brain works is that's the full stack works is that's the full stack works is that's the full stack developer. And I think we're all developer. And I think we're all developer. And I think we're all fumbling around in the dark trying to fumbling around in the dark trying to fumbling around in the dark trying to figure out what's the right way to chain figure out what's the right way to chain figure out what's the right way to chain requests, transform data, automate requests, transform data, automate requests, transform data, automate actions at a point that is specific actions at a point that is specific actions at a point that is specific enough that it's deterministic, enough that it's deterministic, enough that it's deterministic, but also not so specific that it's but also not so specific that it's but also not so specific that it's exhausting.

exhausting. exhausting. >> Yeah, I I think I think you just touched >> Yeah, I I think I think you just touched >> Yeah, I I think I think you just touched on something very powerful. um like the on something very powerful. um like the on something very powerful. um like the visual aspect of programming that we all visual aspect of programming that we all visual aspect of programming that we all do like you know as engineers when do like you know as engineers when do like you know as engineers when you're describing something how often do you're describing something how often do you're describing something how often do you reach for a piece of paper or a you reach for a piece of paper or a you reach for a piece of paper or a napkin you start drawing out boxes and napkin you start drawing out boxes and napkin you start drawing out boxes and arrows that's for to most extent I think arrows that's for to most extent I think arrows that's for to most extent I think that's a common way for us to decompose that's a common way for us to decompose that's a common way for us to decompose a problem and describe a workflow and a problem and describe a workflow and a problem and describe a workflow and when I was looking at postman flows uh when I was looking at postman flows uh when I was looking at postman flows uh that's the thing that resonated for me that's the thing that resonated for me that's the thing that resonated for me it's like visual languages have been it's like visual languages have been it's like visual languages have been around for a while uh like what's around for a while uh like what's around for a while uh like what's different uh okay APIs changed the game different uh okay APIs changed the game different uh okay APIs changed the game quite a bit because Postman, like we quite a bit because Postman, like we quite a bit because Postman, like we said before, all of the world's APIs are said before, all of the world's APIs are said before, all of the world's APIs are on Postman and increasingly software on Postman and increasingly software on Postman and increasingly software development today is just composed a development today is just composed a development today is just composed a bunch of APIs together to get a job bunch of APIs together to get a job bunch of APIs together to get a job done. Um, but in the age of AI, like why done. Um, but in the age of AI, like why done. Um, but in the age of AI, like why can an 11-year-old program and vibe and can an 11-year-old program and vibe and can an 11-year-old program and vibe and application, it's because they can application, it's because they can application, it's because they can visually see that the game do what they visually see that the game do what they visually see that the game do what they wanted and what they intended. Uh, like wanted and what they intended. Uh, like wanted and what they intended. Uh, like the buttons are in the right place, the the buttons are in the right place, the the buttons are in the right place, the screen is in the right shape. um you screen is in the right shape. um you screen is in the right shape. um you know they did a lacrosse game that's the know they did a lacrosse game that's the know they did a lacrosse game that's the game they play and so or the mechanics game they play and so or the mechanics game they play and so or the mechanics right um and so when you're talking right um and so when you're talking right um and so when you're talking about full stack development on back end about full stack development on back end about full stack development on back end uh and I'm not a software developer uh and I'm not a software developer uh and I'm not a software developer right because we're seeing the gates are right because we're seeing the gates are right because we're seeing the gates are falling like anybody could build a full falling like anybody could build a full falling like anybody could build a full stack application today how do I know stack application today how do I know stack application today how do I know how to maintain this code in the back how to maintain this code in the back how to maintain this code in the back end it's not giving me a pile of end it's not giving me a pile of end it's not giving me a pile of JavaScript code because I can't reason JavaScript code because I can't reason JavaScript code because I can't reason about it don't give me a pile of Python about it don't give me a pile of Python about it don't give me a pile of Python code that I might not even know how to code that I might not even know how to code that I might not even know how to run I want a visual workflow that run I want a visual workflow that run I want a visual workflow that matches my mental model for the back matches my mental model for the back matches my mental model for the back end. And to me, that's what's powerful

end. And to me, that's what's powerful end. And to me, that's what's powerful about visual languages like postman about visual languages like postman about visual languages like postman flows. Uh I vibe coded a full stack flows. Uh I vibe coded a full stack flows. Uh I vibe coded a full stack application like DUI. The back end is application like DUI. The back end is application like DUI. The back end is completely done with Postman flows, but completely done with Postman flows, but completely done with Postman flows, but I can visually see, hey, is this doing I can visually see, hey, is this doing I can visually see, hey, is this doing what I want or not? Is the logical order what I want or not? Is the logical order what I want or not? Is the logical order of the do this and that um sort of there of the do this and that um sort of there of the do this and that um sort of there in place. Do I need to change an edge? in place. Do I need to change an edge? in place. Do I need to change an edge? Is it faster for me to go and fix Is it faster for me to go and fix Is it faster for me to go and fix something just by tweaking the visual something just by tweaking the visual something just by tweaking the visual language versus having to interact with language versus having to interact with language versus having to interact with the LLM and give it the right context the LLM and give it the right context the LLM and give it the right context just to tweak small thing? Uh so I think just to tweak small thing? Uh so I think just to tweak small thing? Uh so I think that's what's powerful for me and why that's what's powerful for me and why that's what's powerful for me and why we're seeing more and more visual we're seeing more and more visual we're seeing more and more visual workflow builders today. almost workflow builders today. almost workflow builders today. almost everybody is sort of realizing the same everybody is sort of realizing the same everybody is sort of realizing the same thing at the same time which tells you thing at the same time which tells you thing at the same time which tells you like we're heading in a um into a like we're heading in a um into a like we're heading in a um into a paradigm shift in terms of how we build paradigm shift in terms of how we build paradigm shift in terms of how we build software where you're vibing you're software where you're vibing you're software where you're vibing you're specifying to the LLM to build your specifying to the LLM to build your specifying to the LLM to build your application and it's a lot more visual application and it's a lot more visual application and it's a lot more visual um than in a traditional sense where you um than in a traditional sense where you um than in a traditional sense where you just wrote code for everything. just wrote code for everything. just wrote code for everything. >> Yeah. >> Yeah. >> Yeah. Hey friends, we're going to take a beat Hey friends, we're going to take a beat Hey friends, we're going to take a beat in this episode and pay the bills with in this episode and pay the bills with in this episode and pay the bills with our new sponsor ShareGate by Work Leap.

our new sponsor ShareGate by Work Leap. our new sponsor ShareGate by Work Leap. I've got Shaylen Gimy on the line. help I've got Shaylen Gimy on the line. help I've got Shaylen Gimy on the line. help me understand as companies are preparing me understand as companies are preparing me understand as companies are preparing for the end of life of Microsoft for the end of life of Microsoft for the end of life of Microsoft SharePoint 2016 and 2019. I know there's SharePoint 2016 and 2019. I know there's SharePoint 2016 and 2019. I know there's someone on the show here who's been someone on the show here who's been someone on the show here who's been listening who's got that version of listening who's got that version of listening who's got that version of SharePoint. What's some advice you would SharePoint. What's some advice you would SharePoint. What's some advice you would give them to prepare for that end of give them to prepare for that end of give them to prepare for that end of life? life? life? >> I think the biggest mistake is waiting >> I think the biggest mistake is waiting >> I think the biggest mistake is waiting for leadership to make that call. You for leadership to make that call. You for leadership to make that call. You know, if you know a migration is coming, know, if you know a migration is coming, know, if you know a migration is coming, start assessing and planning now. When start assessing and planning now. When start assessing and planning now. When that request finally comes down the that request finally comes down the that request finally comes down the pipeline and you know it's going to be pipeline and you know it's going to be pipeline and you know it's going to be really quick, you'll have a game plan really quick, you'll have a game plan really quick, you'll have a game plan ready. That's where ShareGate comes in. ready. That's where ShareGate comes in. ready. That's where ShareGate comes in. You know, our pre-migration tools help You know, our pre-migration tools help You know, our pre-migration tools help you analyze your environment, map out you analyze your environment, map out you analyze your environment, map out dependencies, and clean up before the dependencies, and clean up before the dependencies, and clean up before the move. The sooner you get ahead of it, move. The sooner you get ahead of it, move. The sooner you get ahead of it, the smoother the transition will be. And the smoother the transition will be. And the smoother the transition will be. And if you're not sure where to start, talk if you're not sure where to start, talk if you're not sure where to start, talk to our experts. You know, get a demo and to our experts. You know, get a demo and to our experts. You know, get a demo and just see how easy ShareGate makes it. just see how easy ShareGate makes it. just see how easy ShareGate makes it. The clock is ticking and the sooner you The clock is ticking and the sooner you The clock is ticking and the sooner you start, the better prepared you'll be. start, the better prepared you'll be. start, the better prepared you'll be. >> The clock is ticking. Check them out at >> The clock is ticking. Check them out at >> The clock is ticking. Check them out at sharegate.com. Sharegate by workle is sharegate.com. Sharegate by workle is sharegate.com. Sharegate by workle is one tool that'll help you migrate faster one tool that'll help you migrate faster one tool that'll help you migrate faster and secure your tenant out of the box.

and secure your tenant out of the box. and secure your tenant out of the box. Microsoft 365 migration best practices Microsoft 365 migration best practices Microsoft 365 migration best practices that guide you at every step. And thank that guide you at every step. And thank that guide you at every step. And thank you to Sharegate for being a sponsor of you to Sharegate for being a sponsor of you to Sharegate for being a sponsor of Hansel minutes. Check them out. Hansel minutes. Check them out. Hansel minutes. Check them out. sharegate.com. sharegate.com. sharegate.com. Now I I want the audience to understand Now I I want the audience to understand Now I I want the audience to understand a little bit of context here and I want a little bit of context here and I want a little bit of context here and I want to gas you up a little bit which is a to gas you up a little bit which is a to gas you up a little bit which is a good thing as the kids say. To be clear, good thing as the kids say. To be clear, good thing as the kids say. To be clear, you were a research scientist at MIT. you were a research scientist at MIT. you were a research scientist at MIT. You have worked on compiler You have worked on compiler You have worked on compiler optimizations. You have a background in optimizations. You have a background in optimizations. You have a background in distributed systems. You know, you you distributed systems. You know, you you distributed systems. You know, you you you understand this stuff. And as you understand this stuff. And as you understand this stuff. And as someone who has thought about it at the, someone who has thought about it at the, someone who has thought about it at the, for lack of a better term, the lex and for lack of a better term, the lex and for lack of a better term, the lex and yak level of like parsing, you're yak level of like parsing, you're yak level of like parsing, you're preferring and finding yourself floating preferring and finding yourself floating preferring and finding yourself floating into a visual language. Now, is there an into a visual language. Now, is there an into a visual language. Now, is there an a is there a compiler? Or is this just a a is there a compiler? Or is this just a a is there a compiler? Or is this just a fundamentally new way of plugging a UI fundamentally new way of plugging a UI fundamentally new way of plugging a UI with business logic and text boxes over with business logic and text boxes over with business logic and text boxes over data into a system? Is it a compiler or data into a system? Is it a compiler or data into a system? Is it a compiler or is it something entirely new? Yeah, I I is it something entirely new? Yeah, I I is it something entirely new? Yeah, I I haven't heard the word the words lex and haven't heard the word the words lex and haven't heard the word the words lex and yak in a long time. So, thanks for yak in a long time. So, thanks for yak in a long time. So, thanks for bringing me back. You're welcome. Um, bringing me back. You're welcome. Um, bringing me back. You're welcome. Um, there is a compiler. Uh, Postman flows there is a compiler. Uh, Postman flows there is a compiler. Uh, Postman flows is a visual language. Uh, it's a visual is a visual language. Uh, it's a visual is a visual language. Uh, it's a visual programming language. There's a compiler programming language. There's a compiler programming language. There's a compiler underneath that takes that and turns it underneath that takes that and turns it underneath that takes that and turns it into a runtime specification. And into a runtime specification. And into a runtime specification. And there's a a runtime, a virtual machine there's a a runtime, a virtual machine there's a a runtime, a virtual machine that actually can execute that bite that actually can execute that bite that actually can execute that bite code, if you will. Um, but just like no, code, if you will. Um, but just like no, code, if you will. Um, but just like no, nobody thinks about assembly anymore. uh nobody thinks about assembly anymore. uh nobody thinks about assembly anymore. uh unless you're working at that level.

unless you're working at that level. unless you're working at that level. Maybe if you're a gaming developer, Maybe if you're a gaming developer, Maybe if you're a gaming developer, you're still coding at the assembly you're still coding at the assembly you're still coding at the assembly level because every microcond counts. Um level because every microcond counts. Um level because every microcond counts. Um you know, you don't peel the layers of you know, you don't peel the layers of you know, you don't peel the layers of the onion deep enough when you're the onion deep enough when you're the onion deep enough when you're building software today. It's more uh building software today. It's more uh building software today. It's more uh and this is why we're like searching for and this is why we're like searching for and this is why we're like searching for the right level of abstraction is so the right level of abstraction is so the right level of abstraction is so powerful. Um so my my ideal state is powerful. Um so my my ideal state is powerful. Um so my my ideal state is like you're just building an like you're just building an like you're just building an application. Uh there's a lot of assets application. Uh there's a lot of assets application. Uh there's a lot of assets generated for you under the covers and generated for you under the covers and generated for you under the covers and you only need to go and peek in. Uh you only need to go and peek in. Uh you only need to go and peek in. Uh maybe you're very curious, but you need maybe you're very curious, but you need maybe you're very curious, but you need to go fix something and just much easier to go fix something and just much easier to go fix something and just much easier for you to just go look at the visual for you to just go look at the visual for you to just go look at the visual workflow. Tweak an edge, tweak a workflow. Tweak an edge, tweak a workflow. Tweak an edge, tweak a variable, tweak a you know small variable, tweak a you know small variable, tweak a you know small statement in context as opposed to statement in context as opposed to statement in context as opposed to having to understand the semantics of having to understand the semantics of having to understand the semantics of JavaScript or Python or whatever is that JavaScript or Python or whatever is that JavaScript or Python or whatever is that you know the the language that you're you know the the language that you're you know the the language that you're using. Um but you don't want to expose using. Um but you don't want to expose using. Um but you don't want to expose that. Do you feel that people of a that. Do you feel that people of a that. Do you feel that people of a certain age, myself included, and others certain age, myself included, and others certain age, myself included, and others are pushing back on this paradigm shift are pushing back on this paradigm shift are pushing back on this paradigm shift that we're in, out of a sense of that we're in, out of a sense of that we're in, out of a sense of mourning? mourning? mourning? Like you're do you feel like you're Like you're do you feel like you're Like you're do you feel like you're throwing away 20 years of experience for throwing away 20 years of experience for throwing away 20 years of experience for another paradigm?

another paradigm? another paradigm? >> Yeah, they so you're asking me like I >> Yeah, they so you're asking me like I >> Yeah, they so you're asking me like I think that's a very good question. when think that's a very good question. when think that's a very good question. when I was building highle languages for I was building highle languages for I was building highle languages for programming hardware um you know we programming hardware um you know we programming hardware um you know we invented a language very similar to Java invented a language very similar to Java invented a language very similar to Java called liquid metal um and we would called liquid metal um and we would called liquid metal um and we would compile that down to verarilog which is compile that down to verarilog which is compile that down to verarilog which is the way you program hardware it's a the way you program hardware it's a the way you program hardware it's a level of gates and wires literally gates level of gates and wires literally gates level of gates and wires literally gates and wires and wires and wires and we would walk into um you know and we would walk into um you know and we would walk into um you know potential customers to to do demos for potential customers to to do demos for potential customers to to do demos for them and we would show them this high them and we would show them this high them and we would show them this high level program um that maybe did high level program um that maybe did high level program um that maybe did high frequency trading and we would show them frequency trading and we would show them frequency trading and we would show them how we're one click we're essentially how we're one click we're essentially how we're one click we're essentially programming the hardware and up we have programming the hardware and up we have programming the hardware and up we have customized hardware that can perform customized hardware that can perform customized hardware that can perform better very low power very attractive better very low power very attractive better very low power very attractive and they would ask the question can I and they would ask the question can I and they would ask the question can I look at the generated vera log so I can look at the generated vera log so I can look at the generated vera log so I can change it and my question was like why change it and my question was like why change it and my question was like why why would you ever want that um and that why would you ever want that um and that why would you ever want that um and that never resonated and I feel you know never resonated and I feel you know never resonated and I feel you know bring that back to today I think we're bring that back to today I think we're bring that back to today I think we're at the same level like if I'm at the same level like if I'm at the same level like if I'm interacting with an LLM and it's interacting with an LLM and it's interacting with an LLM and it's generating creating my application for generating creating my application for generating creating my application for me. Uh I think you know for us people me. Uh I think you know for us people me. Uh I think you know for us people who have been doing this for a while and who have been doing this for a while and who have been doing this for a while and sort of know how the sausage is made sort of know how the sausage is made sort of know how the sausage is made like you want to look underneath. U but like you want to look underneath. U but like you want to look underneath. U but for a large class of programmers like for a large class of programmers like for a large class of programmers like why I don't think I ever want to look why I don't think I ever want to look why I don't think I ever want to look underneath. So and I think resisting it underneath. So and I think resisting it underneath. So and I think resisting it is not necessarily um maybe it's job is not necessarily um maybe it's job is not necessarily um maybe it's job security. I don't think it's nostalgia.

security. I don't think it's nostalgia. security. I don't think it's nostalgia. I think it's a lack of understanding I think it's a lack of understanding I think it's a lack of understanding that your knowledge translates to how that your knowledge translates to how that your knowledge translates to how you wield this power to make you even you wield this power to make you even you wield this power to make you even much more productive. I think that was much more productive. I think that was much more productive. I think that was the thing for me that was the thing for me that was the thing for me that was transformational. Uh I'm a developer. I transformational. Uh I'm a developer. I transformational. Uh I'm a developer. I can drop into just about any language can drop into just about any language can drop into just about any language out there and just program it and be out there and just program it and be out there and just program it and be very proficient. Um maybe not every very proficient. Um maybe not every very proficient. Um maybe not every language but some of the common popular language but some of the common popular language but some of the common popular languages. Um and there was a there was languages. Um and there was a there was languages. Um and there was a there was a market shift for me where I stopped a market shift for me where I stopped a market shift for me where I stopped doing that. you know, instead of doing that. you know, instead of doing that. you know, instead of reaching for VS Code to write my reaching for VS Code to write my reaching for VS Code to write my application, I started reaching for application, I started reaching for application, I started reaching for Postman flows. Um, I built full stack Postman flows. Um, I built full stack Postman flows. Um, I built full stack applications, do authentication, applications, do authentication, applications, do authentication, integration with Stripe, uh, generation integration with Stripe, uh, generation integration with Stripe, uh, generation of code, agents, etc. All things I could of code, agents, etc. All things I could of code, agents, etc. All things I could do with JavaScript or run a lambda or do with JavaScript or run a lambda or do with JavaScript or run a lambda or whatever. Um, but it was useful for me whatever. Um, but it was useful for me whatever. Um, but it was useful for me particularly as I was trying to figure particularly as I was trying to figure particularly as I was trying to figure out my idea was to just lay it out step out my idea was to just lay it out step out my idea was to just lay it out step by step in a visual way. Uh, and that by step in a visual way. Uh, and that by step in a visual way. Uh, and that was transformational. And I think with was transformational. And I think with was transformational. And I think with generative AI where you're generating generative AI where you're generating generative AI where you're generating applications, I think that's just the applications, I think that's just the applications, I think that's just the way we're going to develop. Um, within way we're going to develop. Um, within way we're going to develop. Um, within my team, I've been encouraging, and this my team, I've been encouraging, and this my team, I've been encouraging, and this is true for all of Postman, we're is true for all of Postman, we're is true for all of Postman, we're encouraging all of our engineers to encouraging all of our engineers to encouraging all of our engineers to figure out how to embrace AI for figure out how to embrace AI for figure out how to embrace AI for themselves to essentially uplevel and themselves to essentially uplevel and themselves to essentially uplevel and upskill themselves. And what's upskill themselves. And what's upskill themselves. And what's remarkable is that your senior remarkable is that your senior remarkable is that your senior engineers, for example, who know more, engineers, for example, who know more, engineers, for example, who know more, know how to wield the power well. Your know how to wield the power well. Your know how to wield the power well. Your junior engineers need coaching because junior engineers need coaching because junior engineers need coaching because they don't know yet what they don't they don't know yet what they don't they don't know yet what they don't know. And so they might not know what know. And so they might not know what know. And so they might not know what good solutions look like yet. And so you good solutions look like yet. And so you good solutions look like yet. And so you need to make sure that you're training need to make sure that you're training need to make sure that you're training them in the right way so that you can

them in the right way so that you can them in the right way so that you can uplevel them. Like you're not going to uplevel them. Like you're not going to uplevel them. Like you're not going to have team of just full senior engineers have team of just full senior engineers have team of just full senior engineers all the time or staff engineers. You all the time or staff engineers. You all the time or staff engineers. You need to be able to train the people who need to be able to train the people who need to be able to train the people who are already doing it so that they can are already doing it so that they can are already doing it so that they can increasingly wield um you know AI in increasingly wield um you know AI in increasingly wield um you know AI in very productive ways. And Postman is not very productive ways. And Postman is not very productive ways. And Postman is not unique in this. Like there's lots of unique in this. Like there's lots of unique in this. Like there's lots of companies that are trying to figure this companies that are trying to figure this companies that are trying to figure this out as well. But I think if you're not out as well. But I think if you're not out as well. But I think if you're not doing it, you're going to be disrupted doing it, you're going to be disrupted doing it, you're going to be disrupted completely. Like you're going to be left completely. Like you're going to be left completely. Like you're going to be left behind. Somebody's going to figure it behind. Somebody's going to figure it behind. Somebody's going to figure it out for you. and then you're it's too out for you. and then you're it's too out for you. and then you're it's too late. late. late. >> Yeah. I've been trying to figure out at >> Yeah. I've been trying to figure out at >> Yeah. I've been trying to figure out at what level do I want to write my own what level do I want to write my own what level do I want to write my own poetry and enjoy my own haiku and at poetry and enjoy my own haiku and at poetry and enjoy my own haiku and at what level do I'm like yeah this is just what level do I'm like yeah this is just what level do I'm like yeah this is just five paragraph essay. This is pros. I five paragraph essay. This is pros. I five paragraph essay. This is pros. I can just generate this garbage. Um, I I can just generate this garbage. Um, I I can just generate this garbage. Um, I I I have been managing the advertisers on I have been managing the advertisers on I have been managing the advertisers on the Hansel Minutes podcast out of a JSON the Hansel Minutes podcast out of a JSON the Hansel Minutes podcast out of a JSON file that sits in Azure storage and file that sits in Azure storage and file that sits in Azure storage and every time a new uh sponsor comes in, I every time a new uh sponsor comes in, I every time a new uh sponsor comes in, I open the JSON file and I manually edit open the JSON file and I manually edit open the JSON file and I manually edit it in VS Code. And I've done this for 15 it in VS Code. And I've done this for 15 it in VS Code. And I've done this for 15 years and that's insane. However, the years and that's insane. However, the years and that's insane. However, the problem was not annoying enough for me problem was not annoying enough for me problem was not annoying enough for me to write an app. And then I went and I to write an app. And then I went and I to write an app. And then I went and I finally vibe coded an app. I've never finally vibe coded an app. I've never finally vibe coded an app. I've never seen the code for it. It now lives seen the code for it. It now lives seen the code for it. It now lives online and I can manage sponsors and online and I can manage sponsors and online and I can manage sponsors and advertisers and headsh shot and stuff. I advertisers and headsh shot and stuff. I advertisers and headsh shot and stuff. I think it was written in Typescript, but think it was written in Typescript, but think it was written in Typescript, but I've never seen it before. And I think I've never seen it before. And I think I've never seen it before. And I think that what happened for me was the the that what happened for me was the the that what happened for me was the the pro the the the the problem was like 10 pro the the the the problem was like 10 pro the the the the problem was like 10 minutes a week. It was irritating. And I minutes a week. It was irritating. And I minutes a week. It was irritating. And I was like, I'm not going to dedicate was like, I'm not going to dedicate was like, I'm not going to dedicate three or four days of coding to solve a three or four days of coding to solve a three or four days of coding to solve a 10-minute problem.

10-minute problem. 10-minute problem. >> But I vibecoded it in 42 minutes, >> But I vibecoded it in 42 minutes, >> But I vibecoded it in 42 minutes, >> right? and it took longer to get it into >> right? and it took longer to get it into >> right? and it took longer to get it into the cloud than it did to Vibe Code. And the cloud than it did to Vibe Code. And the cloud than it did to Vibe Code. And that was a moment for me. So I think I that was a moment for me. So I think I that was a moment for me. So I think I still want to spend time doing bespoke still want to spend time doing bespoke still want to spend time doing bespoke work. Uh but not boring text boxes over work. Uh but not boring text boxes over work. Uh but not boring text boxes over data work. data work. data work. >> Yeah, I I lots of good nuggets in there >> Yeah, I I lots of good nuggets in there >> Yeah, I I lots of good nuggets in there to try to unpack. Uh just because you to try to unpack. Uh just because you to try to unpack. Uh just because you can doesn't mean you always should. Um can doesn't mean you always should. Um can doesn't mean you always should. Um and I think I've seen that again and I think I've seen that again and I think I've seen that again firsthand using this technology. like firsthand using this technology. like firsthand using this technology. like sure I can have cursor generate a bunch sure I can have cursor generate a bunch sure I can have cursor generate a bunch of uh code for me or have it debug a of uh code for me or have it debug a of uh code for me or have it debug a test uh but it doesn't you know I don't test uh but it doesn't you know I don't test uh but it doesn't you know I don't want to abdigate the responsibility of want to abdigate the responsibility of want to abdigate the responsibility of actually thinking at some point and actually thinking at some point and actually thinking at some point and solving a problem potentially faster solving a problem potentially faster solving a problem potentially faster than having the LLM go and try to batch than having the LLM go and try to batch than having the LLM go and try to batch it into submission essentially uh just it into submission essentially uh just it into submission essentially uh just going to take longer um and there are going to take longer um and there are going to take longer um and there are things that just bring us joy like you things that just bring us joy like you things that just bring us joy like you don't want to abdicate that uh sort of don't want to abdicate that uh sort of don't want to abdicate that uh sort of stop doing it that's what you enjoy you stop doing it that's what you enjoy you stop doing it that's what you enjoy you should keep doing it uh but there's lots should keep doing it uh but there's lots should keep doing it uh but there's lots of things that are mundane or tedious or of things that are mundane or tedious or of things that are mundane or tedious or it would just take you longer. So here's it would just take you longer. So here's it would just take you longer. So here's where you want to guide um you know the where you want to guide um you know the where you want to guide um you know the AI to help you do your job better. Uh AI to help you do your job better. Uh AI to help you do your job better. Uh I've been keeping track like almost I've been keeping track like almost I've been keeping track like almost every pull request or feature that I fix every pull request or feature that I fix every pull request or feature that I fix uh you know in in terms of the Postman uh you know in in terms of the Postman uh you know in in terms of the Postman flows product uh I look at how long it flows product uh I look at how long it flows product uh I look at how long it took me to take it from when I open took me to take it from when I open took me to take it from when I open cursor to the pull request is open um cursor to the pull request is open um cursor to the pull request is open um and uh I've been sharing those with the and uh I've been sharing those with the and uh I've been sharing those with the team again to sort of encourage them to team again to sort of encourage them to team again to sort of encourage them to say look some things are just much say look some things are just much say look some things are just much faster to do um and we can close 150

faster to do um and we can close 150 faster to do um and we can close 150 bugs in a month in terms of input from bugs in a month in terms of input from bugs in a month in terms of input from feedback uh that we get from our feedback uh that we get from our feedback uh that we get from our customers rather than say this is going customers rather than say this is going customers rather than say this is going to be three sprints that are going to to be three sprints that are going to to be three sprints that are going to take us maybe half a quarter to do like take us maybe half a quarter to do like take us maybe half a quarter to do like we could do it in a week. In fact, we're we could do it in a week. In fact, we're we could do it in a week. In fact, we're doing it right now. Um and we're doing it right now. Um and we're doing it right now. Um and we're building agents to sort of look we're building agents to sort of look we're building agents to sort of look we're building those agents in Postman on building those agents in Postman on building those agents in Postman on flows uh to essentially say we get a flows uh to essentially say we get a flows uh to essentially say we get a feedback um ticket from a customer or feedback um ticket from a customer or feedback um ticket from a customer or from our internal users. Uh we build from our internal users. Uh we build from our internal users. Uh we build Postman on Postman. So a lot of uh usage Postman on Postman. So a lot of uh usage Postman on Postman. So a lot of uh usage also comes in from our internal user also comes in from our internal user also comes in from our internal user base. Um automatically triage it. Can AI base. Um automatically triage it. Can AI base. Um automatically triage it. Can AI just fix this? if it can open the pull just fix this? if it can open the pull just fix this? if it can open the pull request and then the expert comes in request and then the expert comes in request and then the expert comes in judges is this a good fix or not. Um and judges is this a good fix or not. Um and judges is this a good fix or not. Um and you want to make sure that this is you you want to make sure that this is you you want to make sure that this is you know going back to the original question know going back to the original question know going back to the original question you're asking like tools matter the you're asking like tools matter the you're asking like tools matter the specification matters because you don't specification matters because you don't specification matters because you don't want this to become noise for your want this to become noise for your want this to become noise for your engineers and it becomes things that engineers and it becomes things that engineers and it becomes things that they ignore. they ignore. they ignore. >> Yeah. >> Yeah. >> Yeah. >> Or they look at it as a waste of time. >> Or they look at it as a waste of time. >> Or they look at it as a waste of time. >> Yeah. The thing that is going to be >> Yeah. The thing that is going to be >> Yeah. The thing that is going to be challenging for the junior engineers, challenging for the junior engineers, challenging for the junior engineers, for the early in career engineers that for the early in career engineers that for the early in career engineers that have that will exist that are exiting have that will exist that are exiting have that will exist that are exiting university in the middle of this moment university in the middle of this moment university in the middle of this moment is not just the specification matters, is not just the specification matters, is not just the specification matters, but specificity matters. And the but specificity matters. And the but specificity matters. And the challenge that I'm having as old man who challenge that I'm having as old man who challenge that I'm having as old man who shakes fist at Cloud is that hey, my car shakes fist at Cloud is that hey, my car shakes fist at Cloud is that hey, my car is knocking. Hey, my car, you know, like is knocking. Hey, my car, you know, like is knocking. Hey, my car, you know, like my fridge just died, right? Like I my fridge just died, right? Like I my fridge just died, right? Like I talked to you, we actually rescheduled talked to you, we actually rescheduled talked to you, we actually rescheduled this podcast cuz my fridge died, right?

this podcast cuz my fridge died, right? this podcast cuz my fridge died, right? I can just accept that the fridge type, I can just accept that the fridge type, I can just accept that the fridge type, but I know the condenser gave out. Like but I know the condenser gave out. Like but I know the condenser gave out. Like there's at what level are the early in there's at what level are the early in there's at what level are the early in career people going to go like, oh, this career people going to go like, oh, this career people going to go like, oh, this this this Postman flow didn't work. Oh, this this Postman flow didn't work. Oh, this this Postman flow didn't work. Oh, it's because I didn't understand how to it's because I didn't understand how to it's because I didn't understand how to do this. I needed some conditional logic do this. I needed some conditional logic do this. I needed some conditional logic in FQL or in Typescript. And how many of in FQL or in Typescript. And how many of in FQL or in Typescript. And how many of them will just go, I don't know, buy a them will just go, I don't know, buy a them will just go, I don't know, buy a new fridge. new fridge. new fridge. >> Right. Right. Yeah. FQL is flow's query >> Right. Right. Yeah. FQL is flow's query >> Right. Right. Yeah. FQL is flow's query language. It was a simpler way of language. It was a simpler way of language. It was a simpler way of programming um back when you actually programming um back when you actually programming um back when you actually had to write code um versus TypeScript. had to write code um versus TypeScript. had to write code um versus TypeScript. that appeal to a different persona uh that appeal to a different persona uh that appeal to a different persona uh that maybe wasn't a typical software that maybe wasn't a typical software that maybe wasn't a typical software developer. developer. developer. >> Uh another great question I wanted to >> Uh another great question I wanted to >> Uh another great question I wanted to before in terms of um and this is where before in terms of um and this is where before in terms of um and this is where interaction iteration with the LLM interaction iteration with the LLM interaction iteration with the LLM matters. Uh my son was learning matters. Uh my son was learning matters. Uh my son was learning geometry. He's like he was had an geometry. He's like he was had an geometry. He's like he was had an earlier problem he was doing today. He earlier problem he was doing today. He earlier problem he was doing today. He was like dad I don't know how to solve was like dad I don't know how to solve was like dad I don't know how to solve this. I was like you have immense power this. I was like you have immense power this. I was like you have immense power I didn't have when I was growing up. I didn't have when I was growing up. I didn't have when I was growing up. Pull up chat GPT ask it to explain it to Pull up chat GPT ask it to explain it to Pull up chat GPT ask it to explain it to you. You know we sat together and I saw you. You know we sat together and I saw you. You know we sat together and I saw the prompts that he was giving. it was the prompts that he was giving. it was the prompts that he was giving. it was like I don't understand this give me like I don't understand this give me like I don't understand this give me more examples and that interaction got more examples and that interaction got more examples and that interaction got him to solve a geometry problem where he him to solve a geometry problem where he him to solve a geometry problem where he figured it out on his own uh and I think figured it out on his own uh and I think figured it out on his own uh and I think it's the same thing with figuring out it's the same thing with figuring out it's the same thing with figuring out software this didn't work how do you software this didn't work how do you software this didn't work how do you specify what didn't work uh to what specify what didn't work uh to what specify what didn't work uh to what level of detail level of detail level of detail >> and you will get better outcomes uh >> and you will get better outcomes uh >> and you will get better outcomes uh because the large language model are because the large language model are because the large language model are remarkable at understanding our language remarkable at understanding our language remarkable at understanding our language but you have to give them the right but you have to give them the right but you have to give them the right input for them to act and when you know input for them to act and when you know input for them to act and when you know when you've vied your application I when you've vied your application I when you've vied your application I think that final step is where you think that final step is where you think that final step is where you we we will lose a lot of people like we we will lose a lot of people like we we will lose a lot of people like that's still a impediment that needs to

that's still a impediment that needs to that's still a impediment that needs to fall. Um and you see you know companies fall. Um and you see you know companies fall. Um and you see you know companies like Replet for example trying to solve like Replet for example trying to solve like Replet for example trying to solve how you take idea to cloud automatically how you take idea to cloud automatically how you take idea to cloud automatically and this is also a problem that we've and this is also a problem that we've and this is also a problem that we've solved with Postman flows where you know solved with Postman flows where you know solved with Postman flows where you know you build your visual workflow on the you build your visual workflow on the you build your visual workflow on the canvas in Postman. You click a button uh canvas in Postman. You click a button uh canvas in Postman. You click a button uh or you can ask agent mode which is or you can ask agent mode which is or you can ask agent mode which is Postman's embedded agent to just deploy Postman's embedded agent to just deploy Postman's embedded agent to just deploy and it takes a second. It's basically and it takes a second. It's basically and it takes a second. It's basically instantaneous. Um, and your code that instantaneous. Um, and your code that instantaneous. Um, and your code that you were building in your canvas and you were building in your canvas and you were building in your canvas and running locally is now in the cloud. Um, running locally is now in the cloud. Um, running locally is now in the cloud. Um, and this is what Rodri's cloud looks and this is what Rodri's cloud looks and this is what Rodri's cloud looks like. And so I hope people try it out like. And so I hope people try it out like. And so I hope people try it out and enjoy it and let us know how it and enjoy it and let us know how it and enjoy it and let us know how it works. But you want to solve that last works. But you want to solve that last works. But you want to solve that last barrier because you don't want to make barrier because you don't want to make barrier because you don't want to make it easy to solve the hard problem and it easy to solve the hard problem and it easy to solve the hard problem and then die over the things that are then die over the things that are then die over the things that are completely mechanizable. Whether it's completely mechanizable. Whether it's completely mechanizable. Whether it's Microsoft's cloud, Amazon's cloud, Microsoft's cloud, Amazon's cloud, Microsoft's cloud, Amazon's cloud, Google's cloud, they're all hard to use. Google's cloud, they're all hard to use. Google's cloud, they're all hard to use. Uh but I think this is where AI and Uh but I think this is where AI and Uh but I think this is where AI and mechanizing that whole deployment mechanizing that whole deployment mechanizing that whole deployment process really can come in and and be process really can come in and and be process really can come in and and be brought to bear. brought to bear. brought to bear. >> So is it running in is there a container >> So is it running in is there a container >> So is it running in is there a container runtime or how is it able to be moved up runtime or how is it able to be moved up runtime or how is it able to be moved up into any cloud so easily? into any cloud so easily? into any cloud so easily? >> Yeah, it's a serverless runtime um this >> Yeah, it's a serverless runtime um this >> Yeah, it's a serverless runtime um this is again using some of the technology is again using some of the technology is again using some of the technology I've built in the past uh where uh it's I've built in the past uh where uh it's I've built in the past uh where uh it's isolated actually runs um in web isolated actually runs um in web isolated actually runs um in web assembly and we can ship it basically to assembly and we can ship it basically to assembly and we can ship it basically to run in any cloud environment. Oh, web run in any cloud environment. Oh, web run in any cloud environment. Oh, web assembly. That's another thing. Web assembly. That's another thing. Web assembly. That's another thing. Web Assembly is having its moment. You're Assembly is having its moment. You're Assembly is having its moment. You're taking all of these things that are taking all of these things that are taking all of these things that are having their moment and like serverless having their moment and like serverless having their moment and like serverless and web assembly and like knowledge of and web assembly and like knowledge of and web assembly and like knowledge of scale. I've made the joke in the past scale. I've made the joke in the past scale. I've made the joke in the past that I have a three-page resume, but that I have a three-page resume, but that I have a three-page resume, but page two was scaling web farms. So, I

page two was scaling web farms. So, I page two was scaling web farms. So, I just had to rip page two out of the just had to rip page two out of the just had to rip page two out of the thing cuz now it's just a checkbox. And thing cuz now it's just a checkbox. And thing cuz now it's just a checkbox. And I'm I'm I'm sad about that. But at the I'm I'm I'm sad about that. But at the I'm I'm I'm sad about that. But at the same time, we're standing on top of same time, we're standing on top of same time, we're standing on top of this. Why is it so hard to deploy a this. Why is it so hard to deploy a this. Why is it so hard to deploy a freaking app, freaking app, freaking app, >> right, >> right, >> right, >> that is resilient and serverless and >> that is resilient and serverless and >> that is resilient and serverless and state and stateful, state and stateful, state and stateful, >> right? And I think that that was the >> right? And I think that that was the >> right? And I think that that was the appeal for serverless for me like coming appeal for serverless for me like coming appeal for serverless for me like coming terms of the transition I had where you terms of the transition I had where you terms of the transition I had where you know FPJs were the hardest things you know FPJs were the hardest things you know FPJs were the hardest things you could use. You couldn't just go buy one could use. You couldn't just go buy one could use. You couldn't just go buy one um you know at Best Buy like GPUs, um you know at Best Buy like GPUs, um you know at Best Buy like GPUs, graphics processors which really gave uh graphics processors which really gave uh graphics processors which really gave uh you know a run terms of crypto and you know a run terms of crypto and you know a run terms of crypto and graphics gamings and where we are today. graphics gamings and where we are today. graphics gamings and where we are today. uh but they're really hard to use uh uh but they're really hard to use uh uh but they're really hard to use uh remain niche still remain niche and remain niche still remain niche and remain niche still remain niche and where where you are today with where where you are today with where where you are today with serverless like I didn't have to think serverless like I didn't have to think serverless like I didn't have to think about the hardware uh like we were about the hardware uh like we were about the hardware uh like we were trying to explain to people your code trying to explain to people your code trying to explain to people your code will run faster it's low power great I will run faster it's low power great I will run faster it's low power great I don't really care about the underlying don't really care about the underlying don't really care about the underlying mechanism uh and that was the aha moment mechanism uh and that was the aha moment mechanism uh and that was the aha moment for for me was like it scales for for me was like it scales for for me was like it scales automatically I don't have to do automatically I don't have to do automatically I don't have to do anything that's great sign me up uh I anything that's great sign me up uh I anything that's great sign me up uh I pay for what I use and nothing else sign pay for what I use and nothing else sign pay for what I use and nothing else sign me up so I think that's what shifted me up so I think that's what shifted me up so I think that's what shifted things for me and this is where we are things for me and this is where we are things for me and this is where we are today. It's a confluence of so many today. It's a confluence of so many today. It's a confluence of so many things from APIs to visual workflow things from APIs to visual workflow things from APIs to visual workflow languages to distributed systems. The languages to distributed systems. The languages to distributed systems. The immense power we have with LLMs all immense power we have with LLMs all immense power we have with LLMs all coming together and I think the future coming together and I think the future coming together and I think the future is very exciting. Like if I'm young and is very exciting. Like if I'm young and is very exciting. Like if I'm young and growing up here, it's very exciting growing up here, it's very exciting growing up here, it's very exciting time.

time. time. >> I I just think it's really cool that >> I I just think it's really cool that >> I I just think it's really cool that someone with your background that that someone with your background that that someone with your background that that that worked in FPGAAS is now building a that worked in FPGAAS is now building a that worked in FPGAAS is now building a tool for running AI agents and API tool for running AI agents and API tool for running AI agents and API automations. Like I think that's just I automations. Like I think that's just I automations. Like I think that's just I don't know. I give somehow I respect it don't know. I give somehow I respect it don't know. I give somehow I respect it even more. you're like you like I know even more. you're like you like I know even more. you're like you like I know how the thing works but I'm choosing not how the thing works but I'm choosing not how the thing works but I'm choosing not to and I'm going to work at this layer to and I'm going to work at this layer to and I'm going to work at this layer abstraction because I want to get cool abstraction because I want to get cool abstraction because I want to get cool work done that helps people and I think work done that helps people and I think work done that helps people and I think that's really great. that's really great. that's really great. >> Yeah. This is like stay hungry, stay >> Yeah. This is like stay hungry, stay >> Yeah. This is like stay hungry, stay learning like especially for us like learning like especially for us like learning like especially for us like technology changes especially with AI. I technology changes especially with AI. I technology changes especially with AI. I have never seen things change this have never seen things change this have never seen things change this rapidly. Uh and you know what does the rapidly. Uh and you know what does the rapidly. Uh and you know what does the future look like? You know my I have a future look like? You know my I have a future look like? You know my I have a kid who will be entering college in a kid who will be entering college in a kid who will be entering college in a few years. You want to give them the few years. You want to give them the few years. You want to give them the right tools to think through critical right tools to think through critical right tools to think through critical thinking, know how to problem solve, how thinking, know how to problem solve, how thinking, know how to problem solve, how to decompose large problems, and then to decompose large problems, and then to decompose large problems, and then set them up into the future. It's going set them up into the future. It's going set them up into the future. It's going to look very different than what I grew to look very different than what I grew to look very different than what I grew up in, what you grew up with. And the up in, what you grew up with. And the up in, what you grew up with. And the technology, I think, is just technology, I think, is just technology, I think, is just transforming in so powerful ways. But transforming in so powerful ways. But transforming in so powerful ways. But that's what's exciting about technology, that's what's exciting about technology, that's what's exciting about technology, you know? It's like, why would you want you know? It's like, why would you want you know? It's like, why would you want any other job? Like, you know, if you're any other job? Like, you know, if you're any other job? Like, you know, if you're a grad student or you're entering a grad student or you're entering a grad student or you're entering college, absolutely learn this stuff.

college, absolutely learn this stuff. college, absolutely learn this stuff. Your job is going to be secure. You just Your job is going to be secure. You just Your job is going to be secure. You just have to figure out how to wield all of have to figure out how to wield all of have to figure out how to wield all of this immense power. Absolutely true. this immense power. Absolutely true. this immense power. Absolutely true. Well, thank you so much Rodria for Well, thank you so much Rodria for Well, thank you so much Rodria for chatting with me today. Rodri is the chatting with me today. Rodri is the chatting with me today. Rodri is the head of product Postman Flows. You can head of product Postman Flows. You can head of product Postman Flows. You can check them out at postman.com. And this check them out at postman.com. And this check them out at postman.com. And this has been another episode of Hansel has been another episode of Hansel has been another episode of Hansel Minutes. And we'll see you again next Minutes. And we'll see you again next Minutes. And we'll see you again next week.