Implementing Conversation Memory in my AI App - Claude Code

[snorts] [snorts] >> Let's just check if YouTube is going >> Let's just check if YouTube is going >> Let's just check if YouTube is going online correctly. Yeah. Yeah. All right. So, we're live now All right. So, we're live now All right. So, we're live now on YouTube. on YouTube. on YouTube. And So, we are going to um So, we are going to um work on the following issue. Looks like we're still live. Everything Looks like we're still live. Everything is going well. is going well. is going well. So, the issue So, So, the issue So, So, the issue So, Hello everyone Hello everyone Hello everyone from YouTube.

from YouTube. from YouTube. Let's see if you are tuning in Let's see if you are tuning in Let's see if you are tuning in correctly. >> Okay, looks like that's going fine. >> Okay, looks like that's going fine. Then uh we can start. So, the Then uh we can start. So, the Then uh we can start. So, the application I'm application I'm application I'm is an app AI application that my is an app AI application that my is an app AI application that my students can use to ask questions. students can use to ask questions. students can use to ask questions. So, for example, So, for example, So, for example, should I use Proxmox? should I use Proxmox? should I use Proxmox? It then queries all of the coaching It then queries all of the coaching It then queries all of the coaching calls that I have done, all of my calls that I have done, all of my calls that I have done, all of my YouTube videos, and it gives a YouTube videos, and it gives a YouTube videos, and it gives a answer answer answer based on that.

based on that. based on that. And this is running on a local And this is running on a local And this is running on a local Kubernetes cluster with a GPU running on Kubernetes cluster with a GPU running on Kubernetes cluster with a GPU running on a local model. a local model. a local model. But, the current issue is that let's say But, the current issue is that let's say But, the current issue is that let's say I ask, "Should I use Proxmox for my home I ask, "Should I use Proxmox for my home I ask, "Should I use Proxmox for my home lab?" lab?" lab?" Then it will Then it will Then it will say this career alignment. For example, say this career alignment. For example, say this career alignment. For example, you should set goals for the current job you should set goals for the current job you should set goals for the current job you want to reverse engineer. you want to reverse engineer. you want to reverse engineer. Um So, if I then, for example, ask a So, if I then, for example, ask a follow-up question for this, so follow-up question for this, so follow-up question for this, so can you expand can you expand can you expand on the career on the career on the career alignment bullet points? alignment bullet points? alignment bullet points? Then Then Then it's going to say, "I don't have that in it's going to say, "I don't have that in it's going to say, "I don't have that in the retrieved content." Right? So, it's the retrieved content." Right? So, it's the retrieved content." Right? So, it's basically treating this as a completely basically treating this as a completely basically treating this as a completely new prompt new prompt new prompt rather than taking in the previous rather than taking in the previous rather than taking in the previous answer and then querying the model for answer and then querying the model for answer and then querying the model for that. So, this is actually a pretty big that. So, this is actually a pretty big that. So, this is actually a pretty big flaw in the application in that sense.

flaw in the application in that sense. flaw in the application in that sense. And And And what I would then want to what I would then want to what I would then want to uh this was some of the feedback I uh this was some of the feedback I uh this was some of the feedback I gathered from one of my test users. gathered from one of my test users. gathered from one of my test users. And uh he said he he flagged this and And uh he said he he flagged this and And uh he said he he flagged this and I'm very grateful for that. I'm very grateful for that. I'm very grateful for that. So, um So, um So, um that is what I am that is what I am that is what I am um going to be implementing now. um going to be implementing now. um going to be implementing now. So, that I or at least like I'm going to So, that I or at least like I'm going to So, that I or at least like I'm going to be working on it. Like I'm probably not be working on it. Like I'm probably not be working on it. Like I'm probably not going to be able to implement the whole going to be able to implement the whole going to be able to implement the whole thing right now. thing right now. thing right now. Oh, actually we're in the completely Oh, actually we're in the completely Oh, actually we're in the completely wrong category on YouTube. Um, here we go. Um, here we go. Edit.

Stream settings. Stream settings. It says travel and events. It says travel and events. It says travel and events. That needs to be changed ASAP. Yeah, science and technology. Okay. Yeah, science and technology. Okay. Good. So, that's working. So, the observation was, as I explained, So, the observation was, as I explained, there's no conversational memory. there's no conversational memory. there's no conversational memory. Follow-up questions don't work. Follow-up questions don't work. Follow-up questions don't work. Each prompt is as if it's processed as Each prompt is as if it's processed as Each prompt is as if it's processed as if it's a brand new standalone query.

if it's a brand new standalone query. if it's a brand new standalone query. There's no dialogue state, no pronoun There's no dialogue state, no pronoun There's no dialogue state, no pronoun resolution, and tell me more about the resolution, and tell me more about the resolution, and tell me more about the capability. capability. capability. So, this is then So, this is then So, this is then um the thing that I want to um the thing that I want to um the thing that I want to um um um work on. work on. work on. So, only the last user message is So, only the last user message is So, only the last user message is extracted. Everything before it, prior extracted. Everything before it, prior extracted. Everything before it, prior user turns, user turns, user turns, is discarded. is discarded. is discarded. Yeah? Yeah? Yeah? So, So, So, two concerns addressed independently. two concerns addressed independently. two concerns addressed independently. The last question after unspecified, how The last question after unspecified, how The last question after unspecified, how do I do that on arm? So, before the embedding, ask a fast So, before the embedding, ask a fast model to rewrite the last question into model to rewrite the last question into model to rewrite the last question into a standalone query using prior turns as a standalone query using prior turns as a standalone query using prior turns as context. context. context. Standard rag pattern adds one small LLM Standard rag pattern adds one small LLM Standard rag pattern adds one small LLM call, but massively improves retrieval call, but massively improves retrieval call, but massively improves retrieval on follow-ups. So, that is one thing. Embed the concatenation of the last two Embed the concatenation of the last two three user messages. Could be cheap. three user messages. Could be cheap. three user messages. Could be cheap. Full conversational embedding, embed Full conversational embedding, embed Full conversational embedding, embed everything. Usually worse because topic everything. Usually worse because topic everything. Usually worse because topic drift pollutes the query vector.

drift pollutes the query vector. drift pollutes the query vector. Recommend A. So, the the problem is like when you So, the the problem is like when you create an applica- when you are having a create an applica- when you are having a create an applica- when you are having a conversation history with this, conversation history with this, conversation history with this, like I'm I'm learning this as I go, but like I'm I'm learning this as I go, but like I'm I'm learning this as I go, but how do you then include this in the pre- how do you then include this in the pre- how do you then include this in the pre- in in the whole conversation? It's in in the whole conversation? It's in in the whole conversation? It's actually very interesting. actually very interesting. actually very interesting. A very interesting challenge here. A very interesting challenge here. A very interesting challenge here. Um Pass dialogue history into the Gemma Pass dialogue history into the Gemma prompt after context block, so the model prompt after context block, so the model prompt after context block, so the model sees the actual back and forth sees the actual back and forth sees the actual back and forth structure. structure. structure. Token budget is tight. So, first of all, I want to fire off So, first of all, I want to fire off some deep research on um some deep research on um some deep research on um on this because like I have no idea how on this because like I have no idea how on this because like I have no idea how this is actually be how would this this is actually be how would this this is actually be how would this actually be actually be actually be uh implemented in real life.

uh implemented in real life. uh implemented in real life. So, I'm going to start with with that. So, I'm going to start with with that. So, I'm going to start with with that. Um Let's see. Let's see. Pull Pull Pull pull the CubeCraft AI repo. Next, read. Next, read. Which issue is this? Next, I need you to read issue 93. Next, I need you to read issue 93. Did we do any research on this prior?

Okay, so I'm going to go private for a Okay, so I'm going to go private for a second. Hey Saladin.

Hey Saladin. Yeah, so this this kind of YouTube or Yeah, so this this kind of YouTube or Yeah, so this this kind of YouTube or this kind of home lab feedback, that is this kind of home lab feedback, that is this kind of home lab feedback, that is stuff that I would usually do for my stuff that I would usually do for my stuff that I would usually do for my private students. private students. private students. So So So I would be happy to react on it, but I would be happy to react on it, but I would be happy to react on it, but like this is the kind of mentoring that like this is the kind of mentoring that like this is the kind of mentoring that I that I only really provide for paying I that I only really provide for paying I that I only really provide for paying customers. customers. customers. So I'm I would be very happy to do it, So I'm I would be very happy to do it, So I'm I would be very happy to do it, but I cannot do it but I cannot do it but I cannot do it like I get thousands of people asking like I get thousands of people asking like I get thousands of people asking this, right? So this, right? So this, right? So if um if you want me to do it, then consider if you want me to do it, then consider joining my mentorship program. joining my mentorship program. joining my mentorship program. >> [snorts] Yes, okay.

Yes, okay. Uh testing if my Uh testing if my Uh testing if my recording recording recording transcription is working. All right, so we have done some research All right, so we have done some research based on this. So the issue is um based on this. So the issue is um based on this. So the issue is um it is it is it is So the the the challenge here is clear So the the the challenge here is clear So the the the challenge here is clear then. then. then. But the next thing the next thing then is that I want to get an actual then is that I want to get an actual then is that I want to get an actual pulse of pulse of pulse of how this has been implemented in real how this has been implemented in real how this has been implemented in real life. life. life. So So So can you search can you search can you search with multiple research agents? with multiple research agents? with multiple research agents? Can you Can you Can you first check the first last first check the first last first check the first last 30 to 60 days on X 30 to 60 days on X 30 to 60 days on X if there has have been any announcements if there has have been any announcements if there has have been any announcements around this, any recent developments? around this, any recent developments?

around this, any recent developments? And next I want you to launch And next I want you to launch And next I want you to launch separate research agents with Perplexity separate research agents with Perplexity separate research agents with Perplexity and Axe and Axe and Axe and Grok and Grok and Grok to see what the actual production to see what the actual production to see what the actual production patterns are that are being done by AI patterns are that are being done by AI patterns are that are being done by AI companies or companies that build AI companies or companies that build AI companies or companies that build AI applications applications applications to handle this kind of conversational to handle this kind of conversational to handle this kind of conversational memory. memory. memory. And how what the best practices are to And how what the best practices are to And how what the best practices are to implement this. I have to quickly I have to quickly send a link to one of my friends here.

So, now my um So, now my um my AI is going to launch multiple my AI is going to launch multiple my AI is going to launch multiple research agents. research agents. research agents. Um Um Um Okay. And I have a very elaborate AI Okay. And I have a very elaborate AI Okay. And I have a very elaborate AI agent setup that I've written that I've agent setup that I've written that I've agent setup that I've written that I've built up over many months of use. built up over many months of use. built up over many months of use. And it's now going to define some ideal And it's now going to define some ideal And it's now going to define some ideal state criteria. state criteria. state criteria. As you can see, and it's now going to As you can see, and it's now going to As you can see, and it's now going to fire off multiple research agents to fire off multiple research agents to fire off multiple research agents to um um um research the the things that I just research the the things that I just research the the things that I just requested it to do. In the meantime, what I need to do is to In the meantime, what I need to do is to set up the actual Cube Crafts AI Cube Crafts AI container on my container on my container on my local laptop. Because I believe I don't have that Because I believe I don't have that running here. F 4 LS F 4 LS I do have it.

Yeah, it seems that this one does not Yeah, it seems that this one does not have the env file. Oh, Calvin is here. Hey, Calvin.

Oh, Calvin is here. Hey, Calvin. Appreciate you, man. Appreciate you, man. Appreciate you, man. Had such a good time. What's happening now is that I'm What's happening now is that I'm legends, I have multiple research agents legends, I have multiple research agents legends, I have multiple research agents using Perplexity, using Perplexity, using Perplexity, um um um using Grok research, using Gemini using Grok research, using Gemini using Grok research, using Gemini research, research, research, and then my agent is going to pull in and then my agent is going to pull in and then my agent is going to pull in all of that research, and it's going to all of that research, and it's going to all of that research, and it's going to synthesize the synthesize the synthesize the the the the the correct thing. So, what's powerful the correct thing. So, what's powerful the correct thing. So, what's powerful about this is that this will actually about this is that this will actually about this is that this will actually research on X. research on X. research on X. And like if you're not on X, And like if you're not on X, And like if you're not on X, and you're in AI, then you're truly and you're in AI, then you're truly and you're in AI, then you're truly missing out, because all of the AI missing out, because all of the AI missing out, because all of the AI conversation, conversation, conversation, the latest trends, all of that happens the latest trends, all of that happens the latest trends, all of that happens on X. on X. on X. And Perplexity then is good for And Perplexity then is good for And Perplexity then is good for general research. So, it's I put I I general research. So, it's I put I I general research. So, it's I put I I launch multiple research agents, then I launch multiple research agents, then I launch multiple research agents, then I make multiple AIs talk to each other, make multiple AIs talk to each other, make multiple AIs talk to each other, and then pull it the results back in and then pull it the results back in and then pull it the results back in again. That's one of the advanced again. That's one of the advanced again. That's one of the advanced workflows that I um workflows that I um workflows that I um am planning to teach in the upcoming AI am planning to teach in the upcoming AI am planning to teach in the upcoming AI community.

Yes, Saladin, if you want me to review Yes, Saladin, if you want me to review your project, like that is that is your project, like that is that is your project, like that is that is something that I do for DevOps career something that I do for DevOps career something that I do for DevOps career mentoring, right? And mentoring, right? And mentoring, right? And like like like as much as I like to help as many people as much as I like to help as many people as much as I like to help as many people as possible, like I cannot do that for as possible, like I cannot do that for as possible, like I cannot do that for everyone who just asks me on on a stream everyone who just asks me on on a stream everyone who just asks me on on a stream like this. like this. like this. Um Um Um that is a privilege I reserve for my that is a privilege I reserve for my that is a privilege I reserve for my paying customers and my mentorship paying customers and my mentorship paying customers and my mentorship students. students. students. So, if you would like me to do that, I'm So, if you would like me to do that, I'm So, if you would like me to do that, I'm very happy to do go deep in it, but um very happy to do go deep in it, but um very happy to do go deep in it, but um that is um not something I will do not something I will do publicly like this. publicly like this. publicly like this. That is reserved for my private That is reserved for my private That is reserved for my private mentorship students. mentorship students. mentorship students. Um Uh Uh shoot smoke. shoot smoke. shoot smoke. No.

my dev container here is throwing some my dev container here is throwing some errors when I run it locally. Okay, so now it's working. It's also Okay, so now it's working. It's also loading the end file, so I know now that loading the end file, so I know now that loading the end file, so I know now that I have a I know I have everything the same. I know I have everything the same. Get switch master. Because based on this research, I can Because based on this research, I can then start prompting the Superpowers then start prompting the Superpowers then start prompting the Superpowers agent to start implementing the to start implementing the to start implementing the conversational memory. conversational memory. conversational memory. Yeah, so Aladdin, um if you check out Yeah, so Aladdin, um if you check out Yeah, so Aladdin, um if you check out the the the description below this stream, description below this stream, description below this stream, you are Yeah, you are on YouTube.

you are Yeah, you are on YouTube. you are Yeah, you are on YouTube. So, So, So, actually I can just share it in the actually I can just share it in the actually I can just share it in the chat. chat. chat. Um is it >> Oh, that should actually be {slash} >> Oh, that should actually be {slash} live. Yeah, that's working. Um so, I'll share Yeah, that's working. Um so, I'll share the link here the link here the link here in the the chat here.

Um here, if you go to this link, Um here, if you go to this link, then then then there is the page where you can um there is the page where you can um there is the page where you can um find more information about the find more information about the find more information about the mentorship. mentorship. mentorship. And you can book a free call with my And you can book a free call with my And you can book a free call with my team. team. team. It's completely free. There is no um There is no um And then they will see they will at And then they will see they will at And then they will see they will at first of all explain all the details to first of all explain all the details to first of all explain all the details to you. They will see if you're a good fit you. They will see if you're a good fit you. They will see if you're a good fit and get you on board straight away. So, and get you on board straight away. So, and get you on board straight away. So, you can just book a call there. you can just book a call there. you can just book a call there. And then my team will help you onwards. And then my team will help you onwards. And then my team will help you onwards. But if you have any other questions But if you have any other questions But if you have any other questions about this, just feel free to ask in the about this, just feel free to ask in the about this, just feel free to ask in the in the chat. Are you implementing planning to create Are you implementing planning to create tutorials in school how to implement the tutorials in school how to implement the tutorials in school how to implement the research system? Yes, so research system? Yes, so research system? Yes, so um below this description, I can um below this description, I can um below this description, I can actually share that link as well. actually share that link as well. actually share that link as well. I'm sort of I'm sort of I'm sort of gathering like I've been using AI every gathering like I've been using AI every gathering like I've been using AI every day for almost 2 years now.

day for almost 2 years now. day for almost 2 years now. And I feel I have a lot to teach about And I feel I have a lot to teach about And I feel I have a lot to teach about it, but I am I'm still sort of doubting it, but I am I'm still sort of doubting it, but I am I'm still sort of doubting whether I should start teaching this whether I should start teaching this whether I should start teaching this because I'm not sure if there's enough because I'm not sure if there's enough because I'm not sure if there's enough interest for it. interest for it. interest for it. So, I am considering starting a So, I am considering starting a So, I am considering starting a um um um AI community or an AI uh something AI community or an AI uh something AI community or an AI uh something around AI, around AI, around AI, but I I'm sort of still gathering the but I I'm sort of still gathering the but I I'm sort of still gathering the interest around it, interest around it, interest around it, but since you're asking this question but since you're asking this question but since you're asking this question uh that is good. So, below the uh that is good. So, below the uh that is good. So, below the description, there is already a link description, there is already a link description, there is already a link to a waitlist, to a waitlist, to a waitlist, and I'm going to share that in the chat and I'm going to share that in the chat and I'm going to share that in the chat here as well. here as well. here as well. This is a waitlist for a This is a waitlist for a This is a waitlist for a community community community or for the community that I'm going to or for the community that I'm going to or for the community that I'm going to start. And if you sign up for the start. And if you sign up for the start. And if you sign up for the waitlist, then you are going to get a waitlist, then you are going to get a waitlist, then you are going to get a massive discount for uh when I if I do massive discount for uh when I if I do massive discount for uh when I if I do decide to launch it. decide to launch it. decide to launch it. So, So, So, like in any case, just sign up for the like in any case, just sign up for the like in any case, just sign up for the waitlist because then you're going to waitlist because then you're going to waitlist because then you're going to get the discount. You'll be the first get the discount. You'll be the first get the discount. You'll be the first one to know about it.

one to know about it. one to know about it. Uh but the waitlist is about like if I Uh but the waitlist is about like if I Uh but the waitlist is about like if I can get a few some people on this list, can get a few some people on this list, can get a few some people on this list, then I'll know what they want, right? then I'll know what they want, right? then I'll know what they want, right? So, um So, um So, um yeah, also share in the chat what what yeah, also share in the chat what what yeah, also share in the chat what what are what would you look for in an AI in are what would you look for in an AI in are what would you look for in an AI in AI community? What would you like to AI community? What would you like to AI community? What would you like to learn from me? learn from me? learn from me? And uh let me know in the chat, and join And uh let me know in the chat, and join And uh let me know in the chat, and join the waitlist. And if you do, then you'll the waitlist. And if you do, then you'll the waitlist. And if you do, then you'll be the first one to know about it, and be the first one to know about it, and be the first one to know about it, and you get a big discount. you get a big discount. you get a big discount. Okay, so our research has um Okay, so our research has um Okay, so our research has um has con- Um Um I sup- Flying Superman, I sup- Flying Superman, I sup- Flying Superman, Watching your videos since long, would Watching your videos since long, would Watching your videos since long, would love to learn skills from you by your love to learn skills from you by your love to learn skills from you by your mentorship program. That's cool, man. I mentorship program. That's cool, man. I mentorship program. That's cool, man. I just shared the link in the chat. just shared the link in the chat. just shared the link in the chat. Uh Uh Uh Did you get that one? I can I can share Did you get that one? I can I can share Did you get that one? I can I can share that one again. that one again. that one again. It is the Cube Craft.

It is the Cube Craft. It is the Cube Craft. Uh this is for flying superman. If you want to If you want to Okay, I can pin this message. Flying superman, if you want to check Flying superman, if you want to check that out, this is the link to that. that out, this is the link to that. that out, this is the link to that. Since you're asking about the Since you're asking about the Since you're asking about the mentorship, you can go there mentorship, you can go there mentorship, you can go there and get some info. Let me know if you and get some info. Let me know if you and get some info. Let me know if you have questions about it. have questions about it. have questions about it. Are you in a minivan? Yes, I am in a Are you in a minivan? Yes, I am in a Are you in a minivan? Yes, I am in a minivan. I'm minivan. I'm minivan. I'm building AI apps from the road. And our research has returned. So, the And our research has returned. So, the headline findings on X, so this is why headline findings on X, so this is why headline findings on X, so this is why this is so cool that X is actually that this is so cool that X is actually that this is so cool that X is actually that is where it happens. is where it happens. is where it happens. And there is uh a fresh launch around And there is uh a fresh launch around And there is uh a fresh launch around this with the Cloud Flare agent memory. this with the Cloud Flare agent memory. this with the Cloud Flare agent memory. This course has shifted from rag to This course has shifted from rag to This course has shifted from rag to agentic rag. Okay, that's very cool. So, the implementation primitives So, the implementation primitives haven't changed, okay? Production haven't changed, okay? Production haven't changed, okay? Production consensus on five patterns. consensus on five patterns. consensus on five patterns. A small LLM standalone query rewrite A small LLM standalone query rewrite A small LLM standalone query rewrite pre-retrieval, dialogue history in the pre-retrieval, dialogue history in the pre-retrieval, dialogue history in the prompt with four to eight verbatim prompt with four to eight verbatim prompt with four to eight verbatim terms.

terms. terms. Okay. So, concatenating less and yeah, zero. So, concatenating less and yeah, zero. Okay, so this we have to drop. Even OpenAI is moving customers towards Even OpenAI is moving customers towards app owned conversation state. app owned conversation state. app owned conversation state. Okay. Okay. Okay. So, recommended path ship is issue 93 as So, recommended path ship is issue 93 as So, recommended path ship is issue 93 as designed with the order tweak add designed with the order tweak add designed with the order tweak add rolling summarization next iteration. rolling summarization next iteration. rolling summarization next iteration. The fur line graph until the cross The fur line graph until the cross The fur line graph until the cross section. Okay. section. Okay. section. Okay. So, So, So, if we want to see what this research if we want to see what this research if we want to see what this research actually looks like, like for every actually looks like, like for every actually looks like, like for every query that I launch, query that I launch, query that I launch, the agent is creating a work directory. the agent is creating a work directory. the agent is creating a work directory. So, rag conversation production So, rag conversation production So, rag conversation production patterns. patterns. patterns. So, if I check that out, So, if I check that out, So, if I check that out, uh uh uh here, here, here, and then and then and then CDU memory work, CDU memory work, CDU memory work, and then if I do and then if I do and then if I do rag rag rag patterns, patterns, patterns, then here we can actually see then here we can actually see then here we can actually see all of the research that it has just all of the research that it has just all of the research that it has just done.

done. done. So, this is the synthesis of the So, this is the synthesis of the So, this is the synthesis of the research, but then here are the research, but then here are the research, but then here are the multiple calls that he has done. multiple calls that he has done. multiple calls that he has done. These are These are the actual These are These are the actual These are These are the actual requests that he has done. requests that he has done. requests that he has done. So, it's actually So, it's actually So, it's actually amazing. amazing. amazing. Uh let me see. Uh let me see. Uh let me see. Open AI pattern. Cloud's conversation memory toolkit Cloud's conversation memory toolkit centers on centers on centers on three primitive services in the three primitive services in the three primitive services in the synthesized research prompts caching, context editing, context editing, and hierarchical state files. and hierarchical state files. and hierarchical state files. Interesting. But this then But this then would be a would be a would be a very very very Like that would require to implement Like that would require to implement Like that would require to implement line graph at this stage, which is way line graph at this stage, which is way line graph at this stage, which is way beyond the scope right now.

beyond the scope right now. beyond the scope right now. Multi-turn quality is measured along Multi-turn quality is measured along Multi-turn quality is measured along three axes. Yes, we are not interested in cross Yes, we are not interested in cross conversation memory at this stage. conversation memory at this stage. conversation memory at this stage. So, as he recommends here, then suggest suggest suggest making updates to the issue and suggest making updates to the issue and suggest making updates to the issue and make it more detailed.

make it more detailed. make it more detailed. So, it basically acts acts as a start starter prompt as a start starter prompt as a start starter prompt for the for the for the coding agent to pick it up. coding agent to pick it up. coding agent to pick it up. Now, Yeah, Superman, let me know when you Yeah, Superman, let me know when you have applied. I would um It would be good to know. It would be good to know. Just to see if everything works uh for Just to see if everything works uh for Just to see if everything works uh for you. If the team had some spots you. If the team had some spots you. If the team had some spots available for you. If not, I can pull available for you. If not, I can pull available for you. If not, I can pull some strings if you want to get it some strings if you want to get it some strings if you want to get it earlier.

earlier. earlier. Just let me know, man. Okay, what do Okay, what do It's now showing me what it would update It's now showing me what it would update It's now showing me what it would update on the issue. Yeah, so this is basically sculpting the Yeah, so this is basically sculpting the issue much better.

issue much better. issue much better. So, we want to add the numbers, yeah. Yeah.

Yeah. So, it needs two more things from me. So, it needs two more things from me. So, it needs two more things from me. Separate issue or subtask? I would lean Separate issue or subtask? I would lean Separate issue or subtask? I would lean separate issue. So, um um Please explain. Please explain. Please explain. Yeah. Yeah. Yeah. So, regarding your first question, So, regarding your first question, So, regarding your first question, please explain is the rolling please explain is the rolling please explain is the rolling summarization summarization summarization part of the separate LLM call that we're part of the separate LLM call that we're part of the separate LLM call that we're going to implement going to implement going to implement including the four to eight turns, or is including the four to eight turns, or is including the four to eight turns, or is this something separate entirely? Please this something separate entirely? Please this something separate entirely? Please explain it to me. explain it to me. explain it to me. And then, the second one is we can leave And then, the second one is we can leave And then, the second one is we can leave out these out these out these uh X handles and medium articles, etc.

uh X handles and medium articles, etc. uh X handles and medium articles, etc. That's no um that's not needed in the That's no um that's not needed in the That's no um that's not needed in the actual issue. So, one thing that you will notice if So, one thing that you will notice if you you you see me using see me using see me using Oh, why is it Oh, why is it Oh, why is it Why is it constantly sharing that Linux? Why is it constantly sharing that Linux? Why is it constantly sharing that Linux? That is not That's weird that it show it re That's weird that it show it re My bot is sharing some things. That is My bot is sharing some things. That is My bot is sharing some things. That is not not not Hm. But one thing you'll see me But one thing you'll see me but you notice is that I'm spending a but you notice is that I'm spending a but you notice is that I'm spending a lot of time on research stage on the lot of time on research stage on the lot of time on research stage on the planning phase. planning phase. planning phase. Like the actual writing of the code is Like the actual writing of the code is Like the actual writing of the code is is trivial at this point. Like the real is trivial at this point. Like the real is trivial at this point. Like the real work is here where I'm like asking work is here where I'm like asking work is here where I'm like asking questions refining it and then build questions refining it and then build questions refining it and then build then when the spec is written then the then when the spec is written then the then when the spec is written then the code just writes itself.

code just writes itself. code just writes itself. But I spend multiple hours on But I spend multiple hours on But I spend multiple hours on researching and prompting it properly researching and prompting it properly researching and prompting it properly and that is the real skill of this. and that is the real skill of this. and that is the real skill of this. So there are three different things So there are three different things So there are three different things often conflate. Let me separate them often conflate. Let me separate them often conflate. Let me separate them cleanly. cleanly. cleanly. The rewriter pattern pre-retrieval every The rewriter pattern pre-retrieval every The rewriter pattern pre-retrieval every turn one LM call per turn. turn one LM call per turn. turn one LM call per turn. Condense history and latest user Condense history and latest user Condense history and latest user message. message. message. Okay, so here Okay, so here Okay, so here it would then include like a rewrite in it would then include like a rewrite in it would then include like a rewrite in every call. Imagine turn 12 of a long conversation. Imagine turn 12 of a long conversation. The prompt that goes to Gemma looks like The prompt that goes to Gemma looks like The prompt that goes to Gemma looks like this. System prompt running summary.

this. System prompt running summary. this. System prompt running summary. Yeah. Yeah. Yeah. Forbidden terms. Forbidden terms. Forbidden terms. Rewritten. Rewritten. Rewritten. Yeah. So turn 12 there are three distinct LM So turn 12 there are three distinct LM calls per request at peak. Rewriter calls per request at peak. Rewriter calls per request at peak. Rewriter summarizer and generator. Okay, so we can start with that. Okay, so we can start with that. Phase 1.5 would be at the summarizer Phase 1.5 would be at the summarizer Phase 1.5 would be at the summarizer node. node. node. Triggered when we either A see the real Triggered when we either A see the real Triggered when we either A see the real beta. beta. beta. Okay. Okay, let's update the issue accordingly Okay, let's update the issue accordingly and also include your reasoning for and also include your reasoning for and also include your reasoning for these decisions because we are going to these decisions because we are going to these decisions because we are going to have a second turn with a brainstorming have a second turn with a brainstorming have a second turn with a brainstorming agent. So, that agent needs a little bit agent. So, that agent needs a little bit agent. So, that agent needs a little bit more context than usual that you would more context than usual that you would more context than usual that you would give in issues. So, please update the give in issues. So, please update the give in issues. So, please update the issue now.

Yeah. Yeah. I have an issue with a bot here that I I have an issue with a bot here that I I have an issue with a bot here that I have to turn off. So, let me see have to turn off. So, let me see have to turn off. So, let me see restream. Um Um chat chat chat settings It's constantly sharing this link that It's constantly sharing this link that is not supposed to happen. is not supposed to happen. is not supposed to happen. Reply commands. Reply commands. Reply commands. Edits. Edits. Edits. Save. Uh I use Vim for my IDE. Uh I use Vim for my IDE. Um I'm building an application for my I'm building an application for my private students private students private students that allows them to query all of the that allows them to query all of the that allows them to query all of the all of this all of everything that I've all of this all of everything that I've all of this all of everything that I've I I've 2 years of coaching calls, all of I I've 2 years of coaching calls, all of I I've 2 years of coaching calls, all of my YouTube videos, all of my courses is my YouTube videos, all of my courses is my YouTube videos, all of my courses is embedded in this AI application.

embedded in this AI application. embedded in this AI application. And then when you go into the And then when you go into the And then when you go into the application and ask it a question application and ask it a question application and ask it a question then it asks it will query all of that then it asks it will query all of that then it asks it will query all of that information and then provide an answer information and then provide an answer information and then provide an answer and it also provides the sources and it also provides the sources and it also provides the sources where this was discussed. So, my private where this was discussed. So, my private where this was discussed. So, my private students can go students can go students can go and um there is a whole archive with 2 and um there is a whole archive with 2 and um there is a whole archive with 2 years of coaching calls that I've done years of coaching calls that I've done years of coaching calls that I've done um teaching people how to land DevOps um teaching people how to land DevOps um teaching people how to land DevOps jobs and then they can go and review jobs and then they can go and review jobs and then they can go and review that material as well. that material as well. that material as well. So, that is what uh I'm building. It's So, that is what uh I'm building. It's So, that is what uh I'm building. It's not a public application. This is purely not a public application. This is purely not a public application. This is purely for my private mentorship students. for my private mentorship students. for my private mentorship students. But um But um But um as I'm learning this, I'm going to be as I'm learning this, I'm going to be as I'm learning this, I'm going to be building more public applications as building more public applications as building more public applications as well. well. well. >> [snorts] >> Okay, so the issue is now updated by my >> Okay, so the issue is now updated by my main agent. main agent. main agent. And And And let's see what this looks like now.

A small fast LM takes a dialogue history A small fast LM takes a dialogue history plus the latest user utterance and plus the latest user utterance and plus the latest user utterance and produces a single self-contained search produces a single self-contained search produces a single self-contained search query optimized for vector retrieval. query optimized for vector retrieval. query optimized for vector retrieval. The retrieval layer then embeds the The retrieval layer then embeds the The retrieval layer then embeds the rewritten query. How do I install the Cuba on arm base? How do I install the Cuba on arm base? Yes, exactly. That that looks like a Yes, exactly. That that looks like a Yes, exactly. That that looks like a really good way of doing it. really good way of doing it. really good way of doing it. The model choice, not the The model choice, not the The model choice, not the small not the generator. small not the generator. small not the generator. Um Okay, well, we have to update the issue Okay, well, we have to update the issue to use the same model actually. to use the same model actually. to use the same model actually. Um Due to the infrastructure constraints Due to the infrastructure constraints with only one GPU, I can only run one with only one GPU, I can only run one with only one GPU, I can only run one model at at a time on GPU inference. model at at a time on GPU inference. model at at a time on GPU inference. So, So, So, update the issue so we will actually use update the issue so we will actually use update the issue so we will actually use Gemma for the same uh query at this Gemma for the same uh query at this Gemma for the same uh query at this stage.

>> And then latency budget, that's fine. >> And then latency budget, that's fine. Telemetry. Okay, track to generation dialogue Okay, track to generation dialogue history in a Gemma prompt. The verbatim history in a Gemma prompt. The verbatim history in a Gemma prompt. The verbatim recent dialogue is injected into the recent dialogue is injected into the recent dialogue is injected into the prompt. So, the model sees the actual prompt. So, the model sees the actual prompt. So, the model sees the actual back and forth. Yes. That's good. Okay. That's good. Okay. So, we are almost ready to fire up the So, we are almost ready to fire up the So, we are almost ready to fire up the second agent second agent second agent here.

I'm not so sure about the telemetry I'm not so sure about the telemetry block here though, because that's block here though, because that's block here though, because that's probably going to like we haven't probably going to like we haven't probably going to like we haven't implemented open telemetry yet. And yeah, that link apparently that link And yeah, that link apparently that link doesn't even exist. Uh Uh history.

Why is it taking so long to think about Why is it taking so long to think about this? this? this? There we go. Yes, I have a Yes, I have a about Joseph. about Joseph. about Joseph. I have a full 8-hour course on setting I have a full 8-hour course on setting I have a full 8-hour course on setting up the the development environment. So, up the the development environment. So, up the the development environment. So, everything you see me using now is everything you see me using now is everything you see me using now is running inside of containers, dev running inside of containers, dev running inside of containers, dev containers. containers. containers. Everything is in a containerized Everything is in a containerized Everything is in a containerized workflow. workflow. workflow. And that I teach uh completely in-depth. And that I teach uh completely in-depth. And that I teach uh completely in-depth. And that is already 8 hours of work. And that is already 8 hours of work. And that is already 8 hours of work. And the AI workflows is something that I And the AI workflows is something that I And the AI workflows is something that I am now working on.

am now working on. am now working on. So, if you're interested in that, So, if you're interested in that, So, if you're interested in that, um um um that will come that will come that will come to uh to CubeCraft, but I'm also to uh to CubeCraft, but I'm also to uh to CubeCraft, but I'm also thinking about setting up a separate AI thinking about setting up a separate AI thinking about setting up a separate AI community. community. community. And And And there is a link about yeah, the AI there is a link about yeah, the AI there is a link about yeah, the AI Craftsman. Craftsman. Craftsman. Okay, good. Yeah, Fitzroy just um signed up for the Yeah, Fitzroy just um signed up for the wait list. I can reshare the the wait wait list. I can reshare the the wait wait list. I can reshare the the wait list link for you Joseph. list link for you Joseph. list link for you Joseph. Um that is something that I am going to Um that is something that I am going to Um that is something that I am going to be be be um doing. doing. That I I'm also going to set up the AI That I I'm also going to set up the AI That I I'm also going to set up the AI workflow and AI community possibly when workflow and AI community possibly when workflow and AI community possibly when I have enough interested people I have enough interested people I have enough interested people >> [snorts] >> [snorts] >> [snorts] >> where >> where >> where I will be teaching all the the agent I will be teaching all the the agent I will be teaching all the the agent workflows that you see me using etc. workflows that you see me using etc. workflows that you see me using etc. That's going to be there. That's going to be there. That's going to be there. Uh if you're interested in learning from Uh if you're interested in learning from Uh if you're interested in learning from me about that, then please sign up for me about that, then please sign up for me about that, then please sign up for the wait list and you will also get a the wait list and you will also get a the wait list and you will also get a big discount big discount big discount if I do launch it.

if I do launch it. if I do launch it. So, let me know if you if you sign up So, let me know if you if you sign up So, let me know if you if you sign up for it, that means then it gives me for it, that means then it gives me for it, that means then it gives me a good indication that I'm on the right a good indication that I'm on the right a good indication that I'm on the right track. track. track. And if enough people sign up, then I'll And if enough people sign up, then I'll And if enough people sign up, then I'll I'm I know that I can uh I'm I know that I can uh I'm I know that I can uh it it's going to be a valid thing for me it it's going to be a valid thing for me it it's going to be a valid thing for me to do. Yeah, retriever for the follow-up uses Yeah, retriever for the follow-up uses standalone rewritten query. The rewriter and generator are the same The rewriter and generator are the same Gemma model. Good.

>> Okay. >> Okay. So, at this point, I think I'm ready to So, at this point, I think I'm ready to So, at this point, I think I'm ready to fire off the brainstorming agent for fire off the brainstorming agent for fire off the brainstorming agent for this. So, So, we're on the master branch. I'm going to set this up with remote so I'm going to set this up with remote so that I can also query this from my that I can also query this from my that I can also query this from my phone. phone. phone. Brainstorm. Brainstorm. Brainstorm. No. Okay. Okay. And now I can And now I can And now I can uh uh uh brainstorming.

Let's Let's I have already done a lot of pre-work on I have already done a lot of pre-work on I have already done a lot of pre-work on issue 93, so probably issue 93, so probably issue 93, so probably don't need a very deep brainstorming don't need a very deep brainstorming don't need a very deep brainstorming session. session. session. Read the issue and let's start. >> Yes, right. This is a This is a very >> Yes, right. This is a This is a very privileged position to be in to sit here privileged position to be in to sit here privileged position to be in to sit here with the bird sounds with the bird sounds with the bird sounds and and and uh uh uh sitting here here in my van. sitting here here in my van. sitting here here in my van. And afterwards, I will go for a nice And afterwards, I will go for a nice And afterwards, I will go for a nice bike ride here in Germany. >> So this is now going to

>> So this is now going to launch multiple launch multiple launch multiple explore exploration agents. explore exploration agents. explore exploration agents. And the [snorts] real cool thing with with these with these with this implementation is that it it with this implementation is that it it with this implementation is that it it with superpowers is that it launches sub with superpowers is that it launches sub with superpowers is that it launches sub agents to do everything automatically. I am not sure why this bot keeps I am not sure why this bot keeps um um um saying school.com/linux. saying school.com/linux. saying school.com/linux. I'm sorry about that. Like that is a I'm sorry about that. Like that is a I'm sorry about that. Like that is a it's not very relevant to my it's not very relevant to my it's not very relevant to my current current current stream here. And there is this chat stream here. And there is this chat stream here. And there is this chat setting here. And it has these commands, but And it has these commands, but unfortunately it So is there maybe I think I something I So is there maybe I think I something I can save or something?

can save or something? can save or something? bots You signed up? Oh, that's good. Thank You signed up? Oh, that's good. Thank you. Yeah. Let's see if it will if it's still Yeah. Let's see if it will if it's still working or not. working or not. working or not. Okay, I've read the issue, explored the Okay, I've read the issue, explored the Okay, I've read the issue, explored the code base, you've done the extensive code base, you've done the extensive code base, you've done the extensive pre-work, essentially complete spec pre-work, essentially complete spec pre-work, essentially complete spec already. I have just one clarifying question I have just one clarifying question before proposing approaches. On the before proposing approaches. On the before proposing approaches. On the rewrite of the LM call, the rewriter needs a non-streaming the rewriter needs a non-streaming low-latency call with time max tokens.

low-latency call with time max tokens. low-latency call with time max tokens. Two options: add a small non-streaming Two options: add a small non-streaming Two options: add a small non-streaming helper, Joseph will sign up for Cube Craft in Joseph will sign up for Cube Craft in next couple of months. That's great, next couple of months. That's great, next couple of months. That's great, man. Will you have rolling spots open for Will you have rolling spots open for Cube Craft throughout the year? Cube Craft throughout the year? Cube Craft throughout the year? Well, we are very careful with this Well, we are very careful with this Well, we are very careful with this because like it is a lot of individual because like it is a lot of individual because like it is a lot of individual coaching and attention, so we don't coaching and attention, so we don't coaching and attention, so we don't scale it too fast. scale it too fast. scale it too fast. Uh can I ask you, Joseph, what is Uh can I ask you, Joseph, what is Uh can I ask you, Joseph, what is um what is stopping you from doing it um what is stopping you from doing it um what is stopping you from doing it right now?

right now? right now? I'm very happy to to help you through I'm very happy to to help you through I'm very happy to to help you through this decision here. Good. Good. Design variation is how to organize in Design variation is how to organize in Design variation is how to organize in the code base. >> Yeah, that sounds right. >> Yeah, that sounds right. Yes, we go with A. Okay, so it's now going to propose the Okay, so it's now going to propose the design for the implementation. Take the conversation history plus the Take the conversation history plus the latest user message.

latest user message. latest user message. Call Gemma with a rewriter system prompt Call Gemma with a rewriter system prompt Call Gemma with a rewriter system prompt and return a self-contained search query and return a self-contained search query and return a self-contained search query optimized for vector retrieval. If the history is empty, If the history is empty, yeah, that's correct. >> Um >> Um Well, let's discuss the token limit Well, let's discuss the token limit Well, let's discuss the token limit here. Like, is it really necessary to here. Like, is it really necessary to here. Like, is it really necessary to limit the tokens?

limit the tokens? limit the tokens? Because this is a completely separate Because this is a completely separate Because this is a completely separate LLM call. LLM call. LLM call. Like, theoretically, it could just have Like, theoretically, it could just have Like, theoretically, it could just have as many tokens as many tokens as many tokens as it can handle, right? Since we're not as it can handle, right? Since we're not as it can handle, right? Since we're not actually um adding up all of these tokens. um adding up all of these tokens. It's a completely separate call.

I I get the logic. So, we can we can I I get the logic. So, we can we can limit it, but limit it, but limit it, but >> [sighs] >> I don't think we need to be very >> I don't think we need to be very limiting here. Like, you know this better than me, but Like, you know this better than me, but I think we can be generous with the I think we can be generous with the I think we can be generous with the budget. >> So, the new signature >> So, the new signature question chunks, yeah. These returns get converted into a These returns get converted into a single user message block.

single user message block. single user message block. Yes. Yes. Yes. That looks right. The new flow adds two steps before the The new flow adds two steps before the retrieval. retrieval. retrieval. Replace the current extract question Replace the current extract question Replace the current extract question call with the new helper.

You are You are a platform engineer and you want to get a platform engineer and you want to get a platform engineer and you want to get a fully remote job. That's my goal. a fully remote job. That's my goal. a fully remote job. That's my goal. Yeah, so if you're a platform engineer, Yeah, so if you're a platform engineer, Yeah, so if you're a platform engineer, like like like you're you're going to be able to afford you're you're going to be able to afford you're you're going to be able to afford Kubecraft. Like it's not that expensive. Kubecraft. Like it's not that expensive. Kubecraft. Like it's not that expensive. Like the price is not really the issue. Like the price is not really the issue. Like the price is not really the issue. And And And like I would strongly recommend like I would strongly recommend like I would strongly recommend just just just getting in touch with my team because getting in touch with my team because getting in touch with my team because they're going to help you make your they're going to help you make your they're going to help you make your decision. decision. decision. See if you're the right fit or not. See if you're the right fit or not. See if you're the right fit or not. But if the goal is to get a fully remote But if the goal is to get a fully remote But if the goal is to get a fully remote job like I have now, like I can help you job like I have now, like I can help you job like I have now, like I can help you with this, okay? with this, okay? with this, okay? And And And the the thing is that it's not just about the thing is that it's not just about the tech, you also need to learn how to the tech, you also need to learn how to the tech, you also need to learn how to present yourself, how to convert present yourself, how to convert present yourself, how to convert interviews, how to create a LinkedIn interviews, how to create a LinkedIn interviews, how to create a LinkedIn profile that attracts job offers to you. profile that attracts job offers to you. profile that attracts job offers to you. That's all the kind of stuff that we we That's all the kind of stuff that we we That's all the kind of stuff that we we teach teach teach uh inside of Kubecraft. So, uh inside of Kubecraft. So, uh inside of Kubecraft. So, if there's really no reason for you to if there's really no reason for you to if there's really no reason for you to wait, then wait, then wait, then just click the just do it. Just just click the just do it. Just just click the just do it. Just uh apply and book the call. That's uh apply and book the call. That's uh apply and book the call. That's completely for free. There's no strings completely for free. There's no strings completely for free. There's no strings attached to it.

attached to it. attached to it. And it does And it does And it does um like there's no pressure for you to um like there's no pressure for you to um like there's no pressure for you to join or anything. join or anything. join or anything. But if you do if like the the the the But if you do if like the the the the But if you do if like the the the the the the application is completely for the the application is completely for the the application is completely for free. So, there's no strings attached to free. So, there's no strings attached to free. So, there's no strings attached to it. So, just do it. it. So, just do it. it. So, just do it. Don't wait too long because we might Don't wait too long because we might Don't wait too long because we might stop the enrollment with the if the stop the enrollment with the if the stop the enrollment with the if the group grows too fast. group grows too fast. group grows too fast. >> [snorts] >> Okay, so now it has >> Okay, so now it has created a created a created a design document, which we will review. Stenographical rewrite for retrieval and Stenographical rewrite for retrieval and verbatim.

verbatim. verbatim. Okay, that looks good. Okay, that looks good. Okay, that looks good. New reader by You can see how how elaborate all of You can see how how elaborate all of this is in the specs, right? And this is this is in the specs, right? And this is this is in the specs, right? And this is how how how you get to the flow where you can you get to the flow where you can you get to the flow where you can actually actually actually say, "Fine, let's go."

I hope this I hope this you commit it you commit it you commit it to master. to master. to master. We need to start a new branch and then We need to start a new branch and then We need to start a new branch and then add it there and you can start add it there and you can start add it there and you can start the work. Actually, the sun is coming out. So, as Actually, the sun is coming out. So, as this is cooking, I think I'm going to this is cooking, I think I'm going to this is cooking, I think I'm going to move the setup to the outside. And actually, And actually, like this is just going to cook for for like this is just going to cook for for like this is just going to cook for for a while. So, I'm going to a while. So, I'm going to a while. So, I'm going to quickly go to the restroom quickly go to the restroom quickly go to the restroom as this is running.

as this is running. as this is running. And I'm going to And I'm going to And I'm going to move the camera outside. So, I can sit move the camera outside. So, I can sit move the camera outside. So, I can sit in the sun for a bit. So, I'll be right in the sun for a bit. So, I'll be right in the sun for a bit. So, I'll be right back.

>> I've actually changed my mind. I'm just >> I've actually changed my mind. I'm just going to keep going for going to keep going for going to keep going for maybe half an hour more and then maybe half an hour more and then maybe half an hour more and then I'm going to take a break and do some I'm going to take a break and do some I'm going to take a break and do some cycling anyway. So, cycling anyway. So, cycling anyway. So, let's see. Yep, let's do the sub agent let's see. Yep, let's do the sub agent let's see. Yep, let's do the sub agent driven driven driven approach. So, this is now going to launch multiple So, this is now going to launch multiple sub agents to implement this feature. implement this feature. And when we are live on staging, then And when we are live on staging, then And when we are live on staging, then we should be able to test it out. we should be able to test it out. we should be able to test it out. And that is probably the point where I'm And that is probably the point where I'm And that is probably the point where I'm going to cut off the stream. cut off the stream. >> [snorts] >> [snorts] >> [snorts] >> And we can bring it to production in the >> And we can bring it to production in the >> And we can bring it to production in the next session.

>> Yeah, you're welcome, Joseph. >> Yeah, you're welcome, Joseph. And again, like this is not to pressure you again, like this is not to pressure you or anything. And or anything. And or anything. And I I am not a Like, very often what happens is people Like, very often what happens is people will think one day, one day, one day. will think one day, one day, one day. will think one day, one day, one day. And sometimes they just need a little And sometimes they just need a little And sometimes they just need a little push to get started. And I certainly push to get started. And I certainly push to get started. And I certainly needed that push when I was a nurse and needed that push when I was a nurse and needed that push when I was a nurse and I I I I didn't I was for years I was thinking, I didn't I was for years I was thinking, I didn't I was for years I was thinking, ah, should I do it? Should I not do it? ah, should I do it? Should I not do it? ah, should I do it? Should I not do it? So, that's why I I I say it. So, that's why I I I say it. So, that's why I I I say it. But But But just to show you, Joseph, that just to show you, Joseph, that just to show you, Joseph, that if we go inside if we go inside if we go inside and if I open up the wind section here, and if I open up the wind section here, and if I open up the wind section here, like just in the last days we've landed like just in the last days we've landed like just in the last days we've landed three jobs. three jobs. three jobs. The last 2 weeks we've landed four The last 2 weeks we've landed four The last 2 weeks we've landed four like this is one like this is one like this is one This one was posted yesterday where he This one was posted yesterday where he This one was posted yesterday where he got a job, senior DevOps engineer.

got a job, senior DevOps engineer. got a job, senior DevOps engineer. 3 days ago Ionut landed his job. 3 days ago Ionut landed his job. 3 days ago Ionut landed his job. 3 days ago Dane landed a job. 10 days 3 days ago Dane landed a job. 10 days 3 days ago Dane landed a job. 10 days ago Austin landed a job. ago Austin landed a job. ago Austin landed a job. And Adrian landed a job 6 days ago. And Adrian landed a job 6 days ago. And Adrian landed a job 6 days ago. So like we're landing multiple jobs a So like we're landing multiple jobs a So like we're landing multiple jobs a week. week. week. And Hasnain here, he is a platform And Hasnain here, he is a platform And Hasnain here, he is a platform engineer specialist, okay? engineer specialist, okay? engineer specialist, okay? So we're also helping people go from So we're also helping people go from So we're also helping people go from more media to senior levels as well more media to senior levels as well more media to senior levels as well because of our hard focus on Kubernetes because of our hard focus on Kubernetes because of our hard focus on Kubernetes and and and uh our job application skills. uh our job application skills. uh our job application skills. So the results are there, right? So you So the results are there, right? So you So the results are there, right? So you you you you there's no reason to wait. there's no reason to wait. there's no reason to wait. And um I'm sure that it's going to help you I'm sure that it's going to help you out.

I will apply today. Yeah, that's good, I will apply today. Yeah, that's good, man. Go for it. man. Go for it. man. Go for it. Are you going to get a Framework 13 Pro? Are you going to get a Framework 13 Pro? Are you going to get a Framework 13 Pro? Uh yeah, well, the thing is the Uh yeah, well, the thing is the Uh yeah, well, the thing is the Framework laptop Framework laptop Framework laptop is upgrade, right? So they have an is upgrade, right? So they have an is upgrade, right? So they have an upgrade pack upgrade pack upgrade pack where I can basically take the insides where I can basically take the insides where I can basically take the insides of my Framework and just port it over to of my Framework and just port it over to of my Framework and just port it over to a Framework Pro, which is why I love a Framework Pro, which is why I love a Framework Pro, which is why I love Framework so much as a company that they Framework so much as a company that they Framework so much as a company that they actually allow to do this. actually allow to do this. actually allow to do this. So um as soon as I get the So um as soon as I get the So um as soon as I get the as soon as I get the the notification cuz I signed up for the the notification cuz I signed up for the wait list. wait list. wait list. As soon as I get that notification, I'm As soon as I get that notification, I'm As soon as I get that notification, I'm going to get it. going to get it. going to get it. Like instant buy. I'm just waiting it Like instant buy. I'm just waiting it Like instant buy. I'm just waiting it for it to come in stock. Are these guys starting with no prior Are these guys starting with no prior experience? Uh some of these are like we help people Uh some of these are like we help people like who have no tech experience at all.

like who have no tech experience at all. like who have no tech experience at all. All the way up until senior engineers All the way up until senior engineers All the way up until senior engineers who just want to add 50K to their salary who just want to add 50K to their salary who just want to add 50K to their salary a year. a year. a year. So, we help a broad range of customers. So, we help a broad range of customers. So, we help a broad range of customers. And And And um like one of the guys you saw, he he um like one of the guys you saw, he he um like one of the guys you saw, he he just landed his first junior devops job. just landed his first junior devops job. just landed his first junior devops job. And the other one landed a senior And the other one landed a senior And the other one landed a senior platform engineering role, right? And platform engineering role, right? And platform engineering role, right? And increasing his doubling his salary. increasing his doubling his salary. increasing his doubling his salary. So, we help all of those. Is it possible to get a discount for Is it possible to get a discount for Kubecraft? Like usually not. Kubecraft? Like usually not. Kubecraft? Like usually not. Like the price is already Like the price is already Like the price is already like I I have multiple business mentors like I I have multiple business mentors like I I have multiple business mentors and they're saying, "You're crazy for and they're saying, "You're crazy for and they're saying, "You're crazy for for offering this so cheap for what you for offering this so cheap for what you for offering this so cheap for what you provide and how much individual provide and how much individual provide and how much individual attention you give people." attention you give people." attention you give people." But um my my But um my my But um my my priority is to help people land jobs priority is to help people land jobs priority is to help people land jobs because I struggled so much in the because I struggled so much in the because I struggled so much in the beginning. beginning. beginning. So, that is the main priority for me.

And then usually during these calls And then usually during these calls if you have a particular issue, we can if you have a particular issue, we can if you have a particular issue, we can maybe find a solution to it. maybe find a solution to it. maybe find a solution to it. Team will um assist you on that. Team will um assist you on that. Team will um assist you on that. So, the the application is completely So, the the application is completely So, the the application is completely free. So, I will share the link again. Uh there's no strings attached or Uh there's no strings attached or anything like that. Just go here and then my team will talk Just go here and then my team will talk to you about discounts and that kind of to you about discounts and that kind of to you about discounts and that kind of thing. thing. thing. Again, it's completely free, no strings Again, it's completely free, no strings Again, it's completely free, no strings attached. >> So, the advantage of a agent develop >> So, the advantage of a agent develop agent-driven workflow, as you see me agent-driven workflow, as you see me agent-driven workflow, as you see me using here, is that it's limiting the using here, is that it's limiting the using here, is that it's limiting the context.

context. context. So, every agent gets its own context, so So, every agent gets its own context, so So, every agent gets its own context, so it starts with completely fresh, rather it starts with completely fresh, rather it starts with completely fresh, rather than building up this massive context than building up this massive context than building up this massive context window. Yeah, the battery life is is is decent. Yeah, the battery life is is is decent. Like, I get about 6 5 5 to 6 hours on on Like, I get about 6 5 5 to 6 hours on on Like, I get about 6 5 5 to 6 hours on on a battery a battery a battery for my current one. for my current one. for my current one. So, So, So, um to me that is fine. Like, even when um to me that is fine. Like, even when um to me that is fine. Like, even when I'm traveling with my van like this, I I'm traveling with my van like this, I I'm traveling with my van like this, I will prioritize having electricity, so will prioritize having electricity, so will prioritize having electricity, so that's why I'm on a campsite, right? that's why I'm on a campsite, right? that's why I'm on a campsite, right? And And And um um um but I am really looking forward to doing but I am really looking forward to doing but I am really looking forward to doing the pro upgrade, so that I can have the pro upgrade, so that I can have the pro upgrade, so that I can have better battery life, for sure. better battery life, for sure. better battery life, for sure. Like, with the pro upgrade, it's like 20 Like, with the pro upgrade, it's like 20 Like, with the pro upgrade, it's like 20 hours you get.

>> Hello Mashiko. >> Hello Mashiko. Welcome to the stream.

And today? Good man. And today? Good man. What framework used the arm CPU in that What framework used the arm CPU in that What framework used the arm CPU in that upgrade or just brute force battery? upgrade or just brute force battery? upgrade or just brute force battery? Uh, I actually don't know. Uh, I actually don't know. Uh, I actually don't know. Like I know when you buy the No, I think the upgrade is just going to No, I think the upgrade is just going to be the It's it is the chassis, it's the be the It's it is the chassis, it's the be the It's it is the chassis, it's the battery, the screen, etc. battery, the screen, etc. battery, the screen, etc. So, I have the So, I have the So, I have the I have the Ryzen I have the Ryzen I have the Ryzen AI 300 chipset. AI 300 chipset. AI 300 chipset. And I don't believe that's arm. So, my understanding is that I can just So, my understanding is that I can just take my current chip and everything and take my current chip and everything and take my current chip and everything and just use the upgrade chassis to get the just use the upgrade chassis to get the just use the upgrade chassis to get the battery life and the better screen, etc.

>> What country do you consider the best >> What country do you consider the best country to work remote for? country to work remote for? country to work remote for? I mean I mean I mean like if you can get a remote job for a like if you can get a remote job for a like if you can get a remote job for a US company, that is usually the best. Because US tech jobs have in general the Because US tech jobs have in general the highest salary, right? highest salary, right? highest salary, right? So, I've got friends who work from the So, I've got friends who work from the So, I've got friends who work from the Netherlands for US companies and they're Netherlands for US companies and they're Netherlands for US companies and they're making big money. But, that also has higher competition, But, that also has higher competition, right?

>> How deep should you focus on Ansible on >> How deep should you focus on Ansible on Terraform? Terraform? Terraform? Yeah. Yeah. Yeah. I mean, the answer to that question is I mean, the answer to that question is I mean, the answer to that question is nuanced. It completely depends on where nuanced. It completely depends on where nuanced. It completely depends on where your goal is and what you are your goal is and what you are your goal is and what you are um looking for. um looking for. um looking for. And And And like it is clear that you're looking for like it is clear that you're looking for like it is clear that you're looking for some career mentoring, right? And some career mentoring, right? And some career mentoring, right? And I can provide you with that mentoring. I can provide you with that mentoring. I can provide you with that mentoring. But that does usually mean that we go a But that does usually mean that we go a But that does usually mean that we go a little bit deeper into your situation, little bit deeper into your situation, little bit deeper into your situation, and what your goals are, your previous and what your goals are, your previous and what your goals are, your previous experience, and that is not really experience, and that is not really experience, and that is not really something that I can do on a stream like something that I can do on a stream like something that I can do on a stream like this. this. this. So So So um um um I'm going to share a link, Mashiko. get more information about the get more information about the mentorship that I provide in terms of mentorship that I provide in terms of mentorship that I provide in terms of career.

Working for Swiss comp Let me see. Working for Swiss comp Let me see. Working for Swiss company from Eastern Working for Swiss company from Eastern Working for Swiss company from Eastern Europe. Average salary here is less than Europe. Average salary here is less than Europe. Average salary here is less than 1K euro. 1K euro. 1K euro. Yeah, right. But how What is your average salary when But how What is your average salary when working for the Swiss company? Is that a working for the Swiss company? Is that a working for the Swiss company? Is that a lot higher then? >> Competition would be >> Competition would be killing trying for Dutch US companies killing trying for Dutch US companies killing trying for Dutch US companies like I general booking.

like I general booking. like I general booking. Yeah, those are Yeah, those are Yeah, those are those are big companies. Yeah, we've got those are big companies. Yeah, we've got those are big companies. Yeah, we've got lots of people they're lots of people they're lots of people they're huge devops teams. I have I've had huge devops teams. I have I've had huge devops teams. I have I've had multiple friends who have worked there multiple friends who have worked there multiple friends who have worked there or are working there. Uh where are you trying from Wind Spin Uh where are you trying from Wind Spin Voice? Voice? Voice? Cuz Dutch companies are also good to Cuz Dutch companies are also good to Cuz Dutch companies are also good to work for. They uh work for. They uh work for. They uh usually very international good good usually very international good good usually very international good good English English English and and and uh good salary, too.

I'm going to pre-cool my meal for today, I'm going to pre-cool my meal for today, some sardines. Okay, so it's now finished the upper Okay, so it's now finished the upper implementation. Which I will just quickly look over, but Which I will just quickly look over, but it's probably fine. And then we can do a it's probably fine. And then we can do a it's probably fine. And then we can do a deployment to the test environment. deployment to the test environment. deployment to the test environment. And see if the conversation history is And see if the conversation history is And see if the conversation history is working. >> This should actually be caused by

>> This should actually be caused by pre-commit hooks. pre-commit hooks. pre-commit hooks. Um I have to create an issue for that. I have to create an issue for that. Um Add an issue to the repo to implement Add an issue to the repo to implement pre-commit hooks.

Okay, CI is now passing, so I can merge Okay, CI is now passing, so I can merge it. Do you follow Trends the Traveler? No, I Do you follow Trends the Traveler? No, I don't know that person. don't know that person. don't know that person. Trends the Traveler? Then life gaming pioneer. Okay, that's Then life gaming pioneer. Okay, that's cool. Oh, I think I've seen this guy. Oh, I think I've seen this guy. Yeah, but this is actually Thank you for Yeah, but this is actually Thank you for Yeah, but this is actually Thank you for the tip. I need I need tips on van life the tip. I need I need tips on van life the tip. I need I need tips on van life creators creators creators because I'm basically the devops van because I'm basically the devops van because I'm basically the devops van life pioneer. life pioneer. life pioneer. But But But um um um maybe I I should maybe I should like maybe I I should maybe I should like maybe I I should maybe I should like become this for devops.

become this for devops. become this for devops. Um So, So, what's happening now is that it is he is what's happening now is that it is he is what's happening now is that it is he is going to going to going to create a PR and then it's going to create a PR and then it's going to create a PR and then it's going to deploy this to my local Kubernetes deploy this to my local Kubernetes deploy this to my local Kubernetes cluster. cluster. cluster. And when that is deployed, it is going And when that is deployed, it is going And when that is deployed, it is going to um to um to um we're going to be able to log into the we're going to be able to log into the we're going to be able to log into the staging environment and it should have staging environment and it should have staging environment and it should have the the the um um um new feature implemented. new feature implemented. new feature implemented. Merge it and roll it out. I'm Dutch as well trying to get work at I'm Dutch as well trying to get work at bigger software companies. bigger software companies. bigger software companies. Currently working for Rockstars. Okay, Currently working for Rockstars. Okay, Currently working for Rockstars. Okay, cool. Like if you're if you can maybe get Like if you're if you can maybe get Rockstars to pay for some coaching if Rockstars to pay for some coaching if Rockstars to pay for some coaching if you want if you are interested in you want if you are interested in you want if you are interested in learning more from me.

learning more from me. learning more from me. Like we've had multiple people from the Like we've had multiple people from the Like we've had multiple people from the Netherlands getting their employers to Netherlands getting their employers to Netherlands getting their employers to pay for the course for pay for the course for pay for the course for QBCraft, so. QBCraft, so. QBCraft, so. We're cheap labor. We're cheap labor. We're cheap labor. Four times average salary. Yeah, that's Four times average salary. Yeah, that's Four times average salary. Yeah, that's good. good. good. Lots of Yeah, there's lots of talent in Lots of Yeah, there's lots of talent in Lots of Yeah, there's lots of talent in Eastern Europe. A Turkish wizard doing spells in his A Turkish wizard doing spells in his van. Yes, well van. Yes, well van. Yes, well if there's one thing about this is that if there's one thing about this is that if there's one thing about this is that um um um this like writing code with AI this like writing code with AI this like writing code with AI this is modern magic. So, in some ways this is modern magic. So, in some ways this is modern magic. So, in some ways this is uh this is uh this is uh casting spells. Swiss salaries are insanely high. Okay, Swiss salaries are insanely high. Okay, cool. Yeah, please share more channels when Yeah, please share more channels when you find them. All right, guys. All right, guys. So, now So, now So, now the actual conversational memory is live the actual conversational memory is live the actual conversational memory is live on staging. So, here's the big moment, on staging. So, here's the big moment, on staging. So, here's the big moment, basically.

let's head over to the staging let's head over to the staging environment. Um suggest a few prompts and follow-up Um suggest a few prompts and follow-up questions. Hey, is the staging environment Hey, is the staging environment offline? That's not good. >> So guys, the front end is up. The >> So guys, the front end is up. The CloudFlare tunnel is up.

CloudFlare tunnel is up. CloudFlare tunnel is up. And the rag back end is up. I did I did No, it's a CloudFlare tunnel. Would you call it dark magic?

Would you call it dark magic? Depends on what you do with it. Cool to see you are following Rob's Cool to see you are following Rob's streaming trend. streaming trend. streaming trend. Yes, I Like Rob is my mentor, so I am Yes, I Like Rob is my mentor, so I am Yes, I Like Rob is my mentor, so I am definitely definitely definitely um inspired by him, for sure. I seem to have an issue with the logging I seem to have an issue with the logging into my staging environment, though. So, into my staging environment, though. So, into my staging environment, though. So, please bear with me for a second. I do not understand because I do not understand because prod is working fine and super fast. So, why is working fine and super fast. So, why would staging be would staging be would staging be so slow?

>> I'm having some trouble logging in here. >> I'm having some trouble logging in here. Give me a second.

>> Okay, we're in. Um screen. >> Okay, we're in. Um screen. So, the issue was that my Superbase is So, the issue was that my Superbase is So, the issue was that my Superbase is paused. So, for my staging environment, I'm So, for my staging environment, I'm using a free project. And when you don't use it for a while, And when you don't use it for a while, it will it will it will pause. pause. pause. So, that's why the staging environment So, that's why the staging environment So, that's why the staging environment was down. Not coming up. So, we have to wait until Not coming up. So, we have to wait until that is up, and then we can test our that is up, and then we can test our that is up, and then we can test our staging. staging. staging. >> [sighs] >> [sighs] >> [sighs] >> Um >> Um >> Um I have not tried the Cloudflare D I have not tried the Cloudflare D I have not tried the Cloudflare D ingress controller, but I would rather ingress controller, but I would rather ingress controller, but I would rather wait if they come with an gateway API wait if they come with an gateway API wait if they come with an gateway API implementation.

You're living your dream right now. Yes, You're living your dream right now. Yes, this is the dream, man. this is the dream, man. this is the dream, man. I do not live full-time in this vehicle. I do not live full-time in this vehicle. I do not live full-time in this vehicle. I do have a house or an apartment, I do have a house or an apartment, I do have a house or an apartment, but I am trying to but I am trying to but I am trying to engineer my life so I can, yeah, live engineer my life so I can, yeah, live engineer my life so I can, yeah, live here as long as I can. This is what a get agent according looks This is what a get agent according looks like. Yep. Little did he know that I had already Little did he know that I had already sold his house and moved to Spain for a sold his house and moved to Spain for a sold his house and moved to Spain for a life of drinking and sunny beaches. life of drinking and sunny beaches. life of drinking and sunny beaches. >> [sighs] >> First time joining your stream coding >> First time joining your stream coding from my camera. Yeah, that's cool, man. from my camera. Yeah, that's cool, man. from my camera. Yeah, that's cool, man. From your camper van, that's really From your camper van, that's really From your camper van, that's really cool. Where are you located in your cool. Where are you located in your cool. Where are you located in your right now? I do have plans on getting rid of I do have plans on getting rid of Superbase, but Superbase is Superbase, but Superbase is Superbase, but Superbase is such a good It gives you auth, it gives such a good It gives you auth, it gives such a good It gives you auth, it gives you you you um a database a database everything. So, everything. So, everything. So, when I bring my project live, when I bring my project live, when I bring my project live, I will eventually move it away to my own I will eventually move it away to my own I will eventually move it away to my own self-hosted databases, etc.

self-hosted databases, etc. self-hosted databases, etc. But, for now, it is um definitely the right move so that I can definitely the right move so that I can focus on the actual code and focus on the actual code and focus on the actual code and implementation itself. Why do I live in my car? Because I want Why do I live in my car? Because I want to to to see this kind of scenery, be out in see this kind of scenery, be out in see this kind of scenery, be out in nature, nature, nature, and and and travel while I work. Okay, so our staging is back up again.

Okay, so our staging is back up again. So, I will try to So, I will try to So, I will try to log in We are already logged in. We are already logged in. So, now So, now So, now let's go back to the testing queries. let's go back to the testing queries. let's go back to the testing queries. So, So, So, what does Misha say about getting into what does Misha say about getting into what does Misha say about getting into DevOps? So, this is now querying all of the So, this is now querying all of the knowledge that is knowledge that is knowledge that is in Oh, in Oh, in Oh, bad gateway. Um I will have to see if

Um I will have to see if I need to rotate the key.

>> It's correct in the Adams file. >> It's correct in the Adams file. Check if this is also correct in Check if this is also correct in Check if this is also correct in cluster. And it is literally printing out the And it is literally printing out the secret key to the to the CLI now. So, secret key to the to the CLI now. So, secret key to the to the CLI now. So, I'm I'm glad I'm not showing my screen I'm I'm glad I'm not showing my screen I'm I'm glad I'm not showing my screen now.

Uh I am based in the Netherlands, but I Uh I am based in the Netherlands, but I am currently in Germany. What makes you pick base for planet What makes you pick base for planet scale or trus- Okay, so this is now the output. Okay, so this is now the output. Um Okay. Okay. Let's try it again. New chat. Should I use Proxmox for my New chat. Should I use Proxmox for my home lab?

Okay, so it's working. Okay, so it's working. Then, Then, Then, why does he why does he why does he say that Proxmox say that Proxmox say that Proxmox should not be used. Um Um But, what would he say if I already have But, what would he say if I already have But, what would he say if I already have Proxmox installed? Proxmox installed? Proxmox installed? This is going to be the big test if it This is going to be the big test if it This is going to be the big test if it works. Yep. Nice, it's working. Look. Yep. Nice, it's working. Look. If you already have Proxmox installed, If you already have Proxmox installed, If you already have Proxmox installed, Misha would likely encourage you to Misha would likely encourage you to Misha would likely encourage you to continue using it provided you have a continue using it provided you have a continue using it provided you have a good reason for doing so.

good reason for doing so. good reason for doing so. His advice break That's really good. His advice break That's really good. His advice break That's really good. Look at that. It's working, guys. It's working, guys. Give me Give me Give me a few a few a few test conversations again. Prompts plus test conversations again. Prompts plus test conversations again. Prompts plus follow-up cues. follow-up cues. follow-up cues. It is actually working. Yeah, I was not Yeah, I was not I was not sharing my screen because I am I was not sharing my screen because I am I was not sharing my screen because I am I was working with some credentials, so I was working with some credentials, so I was working with some credentials, so obviously I can't share my screen. Okay, so what does Misha say about Okay, so what does Misha say about getting into DevOps? getting into DevOps? getting into DevOps? Okay. Based on the provided Based on the provided excerpts, Misha offers several excerpts, Misha offers several excerpts, Misha offers several perspectives on getting into DevOps. perspectives on getting into DevOps. perspectives on getting into DevOps. Okay, so this is good. Traineeships.

And now we ask a follow-up question And now we ask a follow-up question about does he recommend any about does he recommend any about does he recommend any certifications to do that? Let's see if certifications to do that? Let's see if certifications to do that? Let's see if this will work. Okay, so that's a good good test. Let's Okay, so that's a good good test. Let's see. How does Misha use Kubernetes in see. How does Misha use Kubernetes in see. How does Misha use Kubernetes in his home lab? >> [laughter] >> [laughter] >> Okay, that is not a good question >> Okay, that is not a good question >> Okay, that is not a good question though. Let's see what happens when I do a a

Let's see what happens when I do a a topic shift. So, what Linux distro do topic shift. So, what Linux distro do topic shift. So, what Linux distro do you think you think you think is best? Let's see. Misha recommends different Let's see. Misha recommends different distributions for learning. distributions for learning. distributions for learning. He states that Arch Linux is the best. He states that Arch Linux is the best. He states that Arch Linux is the best. Okay. Okay. Okay. So, let's say So, let's say So, let's say what why is Arch the best for learning? Okay, so now we'll give a more elaborate Okay, so now we'll give a more elaborate answer. And if I then do a topic shift, And if I then do a topic shift, um um um should should should I get the CKA? But then it would kind of go But then it would kind of go um Okay, so let's let's give this to the AI Okay, so let's let's give this to the AI and then ask what it thinks about this.

Okay, it's good result, working well. Okay, it's good result, working well. First question First question First question answered correctly with grounded answered correctly with grounded answered correctly with grounded sources. sources. sources. Follow up, why is Arch the best for Follow up, why is Arch the best for Follow up, why is Arch the best for learning? learning? learning? The rewriter resolved the implicit The rewriter resolved the implicit The rewriter resolved the implicit reference back to the first answer. And produced a self-contained query. The And produced a self-contained query. The answer's coherent detailed. Okay, so answer's coherent detailed. Okay, so answer's coherent detailed. Okay, so that's actually Can you Can you Everything checks out. Everything checks out. Everything checks out. Okay, so it's doing exactly what it Okay, so it's doing exactly what it Okay, so it's doing exactly what it should. We now have some conversational should. We now have some conversational should. We now have some conversational memory, so we can now promote this to memory, so we can now promote this to memory, so we can now promote this to production. Yes, promote it to prod. Yes, promote it to prod. And when it's running to production, And when it's running to production, And when it's running to production, then I can let my test users know that then I can let my test users know that then I can let my test users know that we are actually live with this.

we are actually live with this. we are actually live with this. And then I'm going to make myself some food, go for a short make myself some food, go for a short bike ride, and chill out a bit before I bike ride, and chill out a bit before I bike ride, and chill out a bit before I have my have my have my mentorship session at 6:00 p.m. Yes, I am using Starlink for the Yes, I am using Starlink for the internet. I have not considered um I have not considered um Open Claw. Open Claw. Open Claw. It feels too mainstream. It feels too mainstream. It feels too mainstream. One thing I am considering though is the One thing I am considering though is the One thing I am considering though is the Hermes agent. but this would be I'm but this would be I'm I'm going to be playing with I'm going to be playing with I'm going to be playing with the Hermes agent.

But as you as you can see, like this is But as you as you can see, like this is all running on a Kubernetes cluster that all running on a Kubernetes cluster that all running on a Kubernetes cluster that is is is self-hosted on my own servers, on my own self-hosted on my own servers, on my own self-hosted on my own servers, on my own GPU. GPU. GPU. So all all of the AI queries you've seen So all all of the AI queries you've seen So all all of the AI queries you've seen me doing, it's all running on my own me doing, it's all running on my own me doing, it's all running on my own local AI hardware. local AI hardware. local AI hardware. And And And as you see, the AI is able to as you see, the AI is able to as you see, the AI is able to deploy new deploy new deploy new completely completely completely the the the whole deployment process is the the the whole deployment process is the the the whole deployment process is completely automated because it's using completely automated because it's using completely automated because it's using flux, it's using um flux, it's using um flux, it's using um it's just simply up updating the tag it's just simply up updating the tag it's just simply up updating the tag image and I've built the whole CICD image and I've built the whole CICD image and I've built the whole CICD infrastructure for this. So, I'm I'm infrastructure for this. So, I'm I'm infrastructure for this. So, I'm I'm really running this project as a proper really running this project as a proper really running this project as a proper DevOps engineer should do. DevOps engineer should do. DevOps engineer should do. And this is all the stuff that I can And this is all the stuff that I can And this is all the stuff that I can teach you teach you teach you um if you're interested in learning from um if you're interested in learning from um if you're interested in learning from me. So, let me know in the chat if me. So, let me know in the chat if me. So, let me know in the chat if like what you're seeing here, if you like what you're seeing here, if you like what you're seeing here, if you like like like GitOps, Kubernetes, deployment with GitOps, Kubernetes, deployment with GitOps, Kubernetes, deployment with flux, all automated with my AI agents. flux, all automated with my AI agents. flux, all automated with my AI agents. Like I teach all of this stuff Like I teach all of this stuff Like I teach all of this stuff step-by-step.

step-by-step. step-by-step. And uh conversation memory is now And uh conversation memory is now And uh conversation memory is now deployed to agent, so check that out. agent, so check that out. But now we can go to my But now we can go to my But now we can go to my application in production. application in production. application in production. And let's see And let's see And let's see if we do if we do if we do um um um Let's do a question. Let's do a question. Let's do a question. Like what would be a good way of asking Like what would be a good way of asking Like what would be a good way of asking a follow-up question? So, a follow-up question? So, a follow-up question? So, uh uh uh Which Linux distro [snorts] should I Which Linux distro [snorts] should I Which Linux distro [snorts] should I use? So, now we should go and elaborate on So, now we should go and elaborate on Ubuntu.

Here. If Yeah, here we go. If you have Here. If Yeah, here we go. If you have no experience with Linux, Misha suggests no experience with Linux, Misha suggests no experience with Linux, Misha suggests the following options. Can you elaborate on the Linux distro we Can you elaborate on the Linux distro we just discussed? just discussed? just discussed? Let's see if it's going to return Ubuntu Let's see if it's going to return Ubuntu Let's see if it's going to return Ubuntu now. Yep, here we go. It's just going to go Yep, here we go. It's just going to go deeper into Ubuntu. Awesome, it's working. Well, we've Awesome, it's working. Well, we've implemented conversational memory implemented conversational memory implemented conversational memory application. application. application. This is amazing. This is amazing. This is amazing. So, what I can do now is I am going into So, what I can do now is I am going into So, what I can do now is I am going into my my my um um um test users directory. Let's see.

Let's see. So, let me share my screen again. So, let me share my screen again. So, let me share my screen again. So, So, So, the channel the channel the channel Hello guys. Based on Geyser's Based on Geyser's feedback, I have now implemented feedback, I have now implemented feedback, I have now implemented conversational memory conversational memory conversational memory and the and the and the chats should now feel more like chats should now feel more like chats should now feel more like conversations rather than conversations rather than conversations rather than one-off one-off one-off prompts. prompts. prompts. I tried it by asking the following I tried it by asking the following I tried it by asking the following questions and it questions and it questions and it picked up the history correctly. if you could if you could give it a give it a give it a spin.

And this is the way you build software, And this is the way you build software, you know, you just start you know, you just start you know, you just start um in my case, it correctly continued in my case, it correctly continued with Ubuntu at this stage. Is the core source based on your notes Is the core source based on your notes or Cube Craft website data? or Cube Craft website data? or Cube Craft website data? Um Um Um it is based on it is based on it is based on the 2 years of Q&A calls we have in Cube the 2 years of Q&A calls we have in Cube the 2 years of Q&A calls we have in Cube Craft, Craft, Craft, all of my YouTube videos, all of my YouTube videos, all of my YouTube videos, and and and I'm soon adding all of the courses as I'm soon adding all of the courses as I'm soon adding all of the courses as well.

This will also This will also have an option have an option have an option to to to select all or some of the sources. All right. All right. It's all working really well. Hermes is way better than OpenClaw. Hermes is way better than OpenClaw. Yeah, it's I've heard good things about Yeah, it's I've heard good things about Yeah, it's I've heard good things about it. Have to go now. Have to go now. Nice to see you, Petter. Thank you for Nice to see you, Petter. Thank you for Nice to see you, Petter. Thank you for tuning in. I'll be switching to Arch Linux once you I'll be switching to Arch Linux once you get the Framework Pro. Yeah, good get the Framework Pro. Yeah, good get the Framework Pro. Yeah, good choice, man. You recommend using Arch first before You recommend using Arch first before going to an containerized approach. I mean, it depends on your skill level, I mean, it depends on your skill level, right? And again, Joseph, like You recommend Arch first. How do you You recommend Arch first. How do you manage your Git workflows to cloud code?

manage your Git workflows to cloud code? manage your Git workflows to cloud code? Yeah. Yeah. Yeah. Like Joseph, you you have so many Like Joseph, you you have so many Like Joseph, you you have so many questions and questions and questions and um um um like you I I'm absolutely sure that I like you I I'm absolutely sure that I like you I I'm absolutely sure that I can help you with with all of this, you can help you with with all of this, you can help you with with all of this, you know. So, know. So, know. So, um I'm going I'm I'm going to have to um I'm going I'm I'm going to have to um I'm going I'm I'm going to have to end the stream now, end the stream now, end the stream now, but but but I'm sure you're it's clear to you now I'm sure you're it's clear to you now I'm sure you're it's clear to you now that I have a lot of things to share, that I have a lot of things to share, that I have a lot of things to share, but I think you would also understand but I think you would also understand but I think you would also understand that I cannot just share everything on that I cannot just share everything on that I cannot just share everything on these live streams and helping people these live streams and helping people these live streams and helping people all the time like this. So, because you all the time like this. So, because you all the time like this. So, because you you like people like you you ask so many you like people like you you ask so many you like people like you you ask so many questions questions questions and um for that I have dedicated created and um for that I have dedicated created and um for that I have dedicated created a dedicated mentorship place where a dedicated mentorship place where a dedicated mentorship place where people can come and uh get the help that people can come and uh get the help that people can come and uh get the help that they they they that based on their situation because I that based on their situation because I that based on their situation because I don't really know your situation right don't really know your situation right don't really know your situation right now, right? So, the advice, for example, now, right? So, the advice, for example, now, right? So, the advice, for example, should you shoot your Arch or the should you shoot your Arch or the should you shoot your Arch or the container-based is very dependent on container-based is very dependent on container-based is very dependent on where you're at. And we would need to where you're at. And we would need to where you're at. And we would need to have a conversation about where you're have a conversation about where you're have a conversation about where you're at in order for me to give a proper um at in order for me to give a proper um at in order for me to give a proper um converse converse converse just tell you this or right now, but just tell you this or right now, but just tell you this or right now, but that might be horrible advice because that might be horrible advice because that might be horrible advice because of where you're at.

I actually prefer to mentor people I actually prefer to mentor people inside CubeCraft inside CubeCraft inside CubeCraft and um and um and um give you the correct advice tailored to give you the correct advice tailored to give you the correct advice tailored to your situation. your situation. your situation. So, that this is if if someone tuned in uh that this is if if someone tuned in uh just now and you they're not aware of uh just now and you they're not aware of uh just now and you they're not aware of uh of this. of this. of this. Um If you If you Like I do provide Like I do provide Like I do provide um um um mentorship mentorship mentorship specifically for these kinds of questions. You can for these kinds of questions. You can find more information about that on this find more information about that on this find more information about that on this link. link. link. And And And if you are interested in um learning if you are interested in um learning if you are interested in um learning more about my AI workflows, I'm going to more about my AI workflows, I'm going to more about my AI workflows, I'm going to be launching an AI community be launching an AI community be launching an AI community which um which um which um if there is enough interest for people if there is enough interest for people if there is enough interest for people Like I'm not sure if I'm going to do it, Like I'm not sure if I'm going to do it, Like I'm not sure if I'm going to do it, but if I get enough people interested, I but if I get enough people interested, I but if I get enough people interested, I will do it.

will do it. will do it. I will also post the the link here. Um Um this is a waitlist for the community. this is a waitlist for the community. this is a waitlist for the community. And if I get enough people to sign up And if I get enough people to sign up And if I get enough people to sign up for it, I will do it. Like right now, for it, I will do it. Like right now, for it, I will do it. Like right now, I'm very time constrained, but if I can I'm very time constrained, but if I can I'm very time constrained, but if I can help a lot of people with these AI help a lot of people with these AI help a lot of people with these AI workflows and I have people interested workflows and I have people interested workflows and I have people interested in it, I will do it. in it, I will do it. in it, I will do it. But But But only if there's enough people only if there's enough people only if there's enough people interested. So, but if you sign up for interested. So, but if you sign up for interested. So, but if you sign up for the waitlist, then you're going to get a the waitlist, then you're going to get a the waitlist, then you're going to get a big discount. Like the the wait list big discount. Like the the wait list big discount. Like the the wait list people are going to get a big discount people are going to get a big discount people are going to get a big discount that the rest of the people won't get if that the rest of the people won't get if that the rest of the people won't get if I do launch it. I do launch it. I do launch it. It can't hurt you just sign up for it, It can't hurt you just sign up for it, It can't hurt you just sign up for it, right? right? right? But if I get enough people interested, I But if I get enough people interested, I But if I get enough people interested, I will will will be creating a community where I share be creating a community where I share be creating a community where I share all of these development workflows, how all of these development workflows, how all of these development workflows, how I do it, I do it, I do it, uh all of that. uh all of that. uh all of that. So, um yeah, you can get access to that So, um yeah, you can get access to that So, um yeah, you can get access to that but just signing up for the wait list. but just signing up for the wait list. but just signing up for the wait list. That's an honest day's work. Yes, it has That's an honest day's work. Yes, it has That's an honest day's work. Yes, it has been an honest day's work. And the AI been an honest day's work. And the AI been an honest day's work. And the AI did great. But this is a great feature did great. But this is a great feature did great. But this is a great feature that was implemented now. I'm super that was implemented now. I'm super that was implemented now. I'm super happy with the result.

happy with the result. happy with the result. And now I'm going to have myself some And now I'm going to have myself some And now I'm going to have myself some food, food, food, get the bike, do some uh movement of the get the bike, do some uh movement of the get the bike, do some uh movement of the body. body. body. And uh at 6:00 p.m. I'm going to have And uh at 6:00 p.m. I'm going to have And uh at 6:00 p.m. I'm going to have some private coaching here in Cube some private coaching here in Cube some private coaching here in Cube Craft. Craft. Craft. So, um So, um So, um I'm uh it's going to be an honest day's I'm uh it's going to be an honest day's I'm uh it's going to be an honest day's work. work. work. Guys, thank you so much for tuning in. Guys, thank you so much for tuning in. Guys, thank you so much for tuning in. I've had a good time streaming. I'm Over I've had a good time streaming. I'm Over I've had a good time streaming. I'm Over the next few days it's going to rain the next few days it's going to rain the next few days it's going to rain here in Germany. here in Germany. here in Germany. So, there's going to be a lot more work So, there's going to be a lot more work So, there's going to be a lot more work to being done on the AI project. So, to being done on the AI project. So, to being done on the AI project. So, make sure you're subscribed to YouTube make sure you're subscribed to YouTube make sure you're subscribed to YouTube and on Twitch. and on Twitch. and on Twitch. On Twitch I stream more of the, you On Twitch I stream more of the, you On Twitch I stream more of the, you know, walk- walking, hiking, coffee know, walk- walking, hiking, coffee know, walk- walking, hiking, coffee drinking chats. So, that's more like IRL drinking chats. So, that's more like IRL drinking chats. So, that's more like IRL streaming. On YouTube it is more streaming. On YouTube it is more streaming. On YouTube it is more tailored to this kind of AI building uh tailored to this kind of AI building uh tailored to this kind of AI building uh streaming. So, if you want to see streaming. So, if you want to see streaming. So, if you want to see everything, check everything, check everything, check Twitch channel. Twitch channel. Twitch channel. Uh I'll share the link to that as well.

That's where I like stream constantly. That's where I like stream constantly. And make sure you're subscribed to that. And make sure you're subscribed to that. And then on in the next few days we're going then on in the next few days we're going then on in the next few days we're going to be building more and more features on to be building more and more features on to be building more and more features on the AI application. the AI application. the AI application. So, thank you so much for tuning in, So, thank you so much for tuning in, So, thank you so much for tuning in, guys. I'm going to end the stream.