# 🐣│cohort-voice
- Session: 37b47599-5af3-4511-b789-30d2475645a3
- Channel: Discord #🐣│cohort-voice
- Started: 2026-06-17T17:00:21.249Z
- Ended: 2026-06-17T17:44:36.582Z
- Participants: duckanbro, takekek, andrej, ECWireless, samkuhlmann, Aphilos • Pharo
[00:00:00] [andrej]: And then the internet. Yeah, I mean it's it's like it's bad. Um I mean it was interesting. The first couple of days I left everyone had pretty high energy and then um this semi-fighting happening that I just like slowly more and more tired. Um yesterday people are chatting out, today they're still working up again. Um it's very interesting. I haven't been like super deep on the conferences, I was just usually like I was to uh focus on one once just getting people and then um uh yeah. Just actually like we can make a new friend is like it's always the most fun thing to do. Um everything else is like feels a little bit like background noise. But did you know anyone this year uh hosting an event on Saturday to join and post a link. Nice. Yeah, I am Andrew Erwin. Uh um uh I have found a deep work, which is uh product design studio since 2018. We've designed a lot of interface and products that were now in the kind of fastering the DeFi ecosystem and um now building another project, um that I'm not totally sure about um yeah, mostly um um coordinating humans in uh uh digital and physical space and um creating uh how do you call organizations. Uh it uses a lot of AI as as part of its um like foundational substitute, so like it would be work without AI. But it's not it's not like I'm not inventing AI. I'm just I'm just using it as a um I have I I mean I've been learning about AI um a long time ago I did the Android uh machine learning course and just linked them out and um just kind of like saw that as a potential to to stick around with and kind of pretty much consistently like leveraged all kinds of AI like enhanced um tools to improve my workflow.
[00:00:00] [takekek]: Yeah, thanks. Thanks, Andre. Thanks, Doug, for hosting. There was a point that earlier where kind of Andre, you were talking about how you were getting like different opinions and that you were using AI to like automate that process a little bit better. And then I think the counterpoint that DK made was like, well, sometimes it's like we just want a case maker to like run with it. So I'm kind of curious if we can like pull on that thread further, because I I can kind of I have a very specific example in my head, but I wanted to know how you were um consulting that. Thanks, Andre. Enjoy. Cheers.
[00:00:00] [duckanbro]: How you doing? Oh nice, yeah. How's that going? Is it uh some good vibes? Go back at vibes. Yeah. Yeah. Yeah, the uh It's been a while since I've been to the conference uh and I do go through a recharge every once in a while. Um we'll try to spread the world a little bit. So yeah, it's um just a little bit on what we're doing right now. We're kind of um just reaching out to a lot of different people, uh trying to get different perspectives on how they're using AI and how their profession in general is changing, how they're adapting to it, you know. Um it's like affecting everyone in different ways and everyone's has different workflows and tools, so we're just kinda kinda it's almost like a little market research to understand how people are going about things. Um so yeah, it's it's real chill and casual, but I have a a list of questions and some of it sounds a little bit odd as I try to get to them. Um try to try to go for about 30 minutes. Alright. Cool. So uh well first off, for the sake we are recording, but for the sake of the recording, can you introduce yourself? Right. Is it what you're building now? Is it uh related or Jason? Right. Yeah, I'd like to get into that a little more. Um Yeah, so you know, you work with a lot of people in the kind of this design space and building space for a while now. How do you see um you know AI coming into your workflows and what you do and your process? Um has it changed dramatically, like kinda how you see in those codes or is it a little less so doing with like face-to-face interactions, you know? Right. Right. I don't know how I get them, but um yeah, so I mean I'm not set too, uh uh one of the issues that I run into is that almost like uh pursue too much like expansion into different directions and it's almost like you need that facilitator even more. You know, it's a lot into that symbol.
[00:03:00] [andrej]: Um I would say um yeah, I'm gonna find it simple with no specifics. So the first time I used flying very specifically in time for some work was around the time when GPT came out and we were running interviews and we wanted to just summarize the whole text that became from interviewing people. So we have sticky notes um and then we stepped them into I they gave us a little like shitty draft and then we reload it and that saves a bit of time. Um writing was a little bit easier because we could um provide some outlines that were more um uh like coherent or get some like some basic kind of um machine written feedback on on text uh and consistency. Um and I have obviously like generated images a little bit, but more like to just demonstrate something or just put something out there and let's have an idea and like in uh a visual specific. Um I think where it has significantly improved the score focus of what concerns my work or my interest is when translating and merging the information that people exchange in order to get aligned. So um a very specific example is if you if you run a branding workshop and you want to brand your organization, you have like five, maybe ten people who all have a different perception of what this organization looks like, they have different points of view. And in the past, uh like the workshops that I've done in the past, we would have personal, and everybody would like this is all the things that should be, this is the values, this is the mission, and it would take a lot of work to just get people to align on it and to like discuss like, hey, is this a generic thing from like can I translate what we're thinking about we're thinking? Because everyone has words and that it exists. And um I have when using it basically they're saying like, let's just save all these snippets, or I like s snap them all together and create a simple statement out of it and um it would be usually very consistent, so you don't actually have to have these long conversations anymore, which I find very nice. Because everybody was like, Oh yeah, I see myself doing that. Um and it makes sense because it's just landing different words together and like maintaining uh sense of the individual things. So I think there's a lot of yeah. Yes, you can actually focus. Uh yes, um, but not I mean I personally don't do too much with Nemesis uh agents per se. Although uh yeah, um personally I use mostly exact now uh file for like sense making and text
[00:03:00] [duckanbro]: Are you guys mainly using uh like um the the chat uh uh uh open AI chat GPT is generation? Are you doing any like Egyptian sound that you like getting um into like clo uh cloud cover uh open cloud or the harnesses, things like that? Automation. Yeah. Have you had any work with like new or design software? You know about some AI integrations. Yeah, that's interesting that that's something that uh I haven't really cut uh about yet. But on the code side, uh sorry are you using codex and and club code a little bit to sort of design out prototypes? And code'em out as HTML and web pages, that's cool. Yeah, so speaking of that, do you have when you know especially when you're working in large groups just sort of uh information out of there? How are you managing uh uh that kind of complex context and having like memory and then um are you are you using different data models and in the Mac like you're saying, or how is there one like uploading files and and letting uh whatever system you're using, like still into something, or um do you have other methods? Yeah. Do you are you having like some overlap between uh like multiple projects or other multiple sessions that you guys are doing where you're being able to kind of say like well I don't know what like look and see if there's anything similar we talked about in this other session kind of kid? Uh yeah, that's uh that's a real cool. Even in my chat sometimes, it's almost like a run of more uh like I'll I'll ask your question, I'll be like, oh yeah, because you're working regular on this, uh this is probably what you're talking about. I was like, no, actually this is something totally separate. I don't want you to be like thinking about that bit. Uh but sometimes yeah, that's really helpful, so uh it's interesting. Um the way you're using in the run that's like uh matchmaker or something uh or you know like um a profile. Um like a Zoom out of view of a bunch of tripods is that's kind of interesting. What kind of things uh I mean so have you noticed uh other changes with new like people and how you how you're working with them? Um uh besides I I I guess everyone has like some level of AI now. Um they're using chat all the time, but I don't know, is it generally people are thinking is changed?
[00:06:00] [andrej]: And um and sometimes Gemini uh for just like to replace people. Um just to search things. Uh in terms of things I I prefer. So like I I would usually my workflow is that I have an idea for something that could be a feature or a piece of infrastructure, and then I would ask Claude to tell me how to do that. Claude explains it. I copy paste it the prompt in Cloud Code or Codex and build a version. Uh should it test it, see if it works, and then um refine it from there and then try and like use it that way. And so in that process I would also maybe sometimes put together a skill or an agent and um make it part of the piece of software. I personally don't I don't use the one, um but um Dan uh one of my friends and designer um he is using um I think Figma MCP I was able to squad code I think. Uh this legitimate I forgot what uh platform he was using. Yeah. Pretty much I mean I I have like I host on GitHub and call fair and uh I like one more complicated piece that has database um I mean I think I don't have a database, but I think it's more sophisticated with with its own data model. Yeah, I would say I'm not knowledgeable on on that just like from a technical standpoint that I can teach you the hardest way to I mean it's not hard best, it's just like uh not not expert right, but um the the tool I built was like uh more complex one I custom data model but I uh use then just create a front end for it and then I basically feed a transcript into it and then I have an agent that has that transcript and fits it into the data model so that I can then show the pieces of data um that use of them. So in like um as a concrete example it would be a conversation uh a uh an online session, people talk about things and do the facilitation to um kind of um extract certain bits of information that people possess and some of that information I want to kind of visualize. So for example, you're working on a project that's your goal in some project uh well, who makes master and have you just achieved? Or who is who would you like to work with? And so like we only have this conversation, that's just information that kind of like floats in the air and uh
[00:06:00] [duckanbro]: I guess about like seeing the Zoom down picture and putting all the pieces together. Have you noticed that changes to that world? Yeah. Types and digital common Have you have you have like push track or have you felt any like issues or you know, just a man of stuff and um you know wading through all that and uh you know it just seems like no more editing, which sometimes is fine, sometimes I don't know. Yeah. Yeah. That was model for all I I get the issues with context switching too because it's like, oh well now I can work on ten things. But then it's like, uh you know like keeping all that inside of uh this like DI now and like managing it all. Um do you wanna talk a little bit about the project router mode? Yeah, I mean uh that sounds like really aligned with kind of the work job you do that's coming on right now too. Um we kinda landed on this like digital co-working space idea where you know basically, you know, we can come together, we can work together, help each other out, educate each other, we're gonna share space, you know, and have something on the land number. So you seem that as you you're kind of almost like a you know, you mentioned that you were kinda seen duration and guidance to these folks that are doing uh tokens on specific problems, specific product. Um almost like a a publisher or something for example. Um is that kind of a model that you're shooting for with this and and then how are you actually like since we're up there so that's the kind of tech layer that you once you're coming together for this? Yeah. Okay, so you you have like a labor so you know you're building out the notes and and different things that you're gonna and then Yeah, okay. Yeah, really we've yeah, I mean like a lot of uh interesting number, which makes sense, you know, or coming from the uh kind of uh design agency area.
[00:09:00] [andrej]: to make sure that there's a letter with that so that we can make more out of that like passive flowing uh piece of knowledge and um make it actually that's possible and show kind of like showing them on a dashboard uh sending them on a telegram chat and um helping people making connections uh asymptomacy after that conversation is over um so yeah that would be incredibly I don't do it for that directly actually no so one example um maybe it could be because there's like people who come back and some of them know each other, others don't maybe I just met like personally but we don't know each other. So um one call uh we had I basically prepared some of the information I have from them from my data make base about these people and what they have done and I basically bring that into the session and it was like here's this person, that's what they're doing. Here's this person, that's what they're doing. So I had that and then put in that same round everyone kind of like send it away the corner and it was a fine and correct what we have said about them. Yeah Yeah I think to some extent that some people are freaking out more just um they're trying to catch up with the CFAS like development and a bit of a bit frantic. Uh other people have just transformed their telephone and say I just have to the urgent best of the time like I would I would basically try to make the sets of agent messages I don't have to do as a person and then exactly um I don't find uh when people just use a send snake or to um uh find information really quickly like something that would previously be a search or like a lambda thing it's just like oh I'm just gonna talk to them around so and uh like have kinds of smoke taking um and sense making as I was I think this I better touch it and that and also generally they um into that um kind of like writing things down and making sense to that yeah I think that I mean I'm not sure if it has to do with a it could be a problem with a capacity thing uh and
[00:09:00] [duckanbro]: So yeah, we're getting some of the things we're trying to keep like knowledge together and trying to set up some workflows, automations that are trying to do the still things and the connections and write some up and straps or as curation kind of elements so it's it is it's still pretty tricky. Even if the model's being so good it's just it's almost like too much at that point. You know how like uh a Discord server is just like or telegrams, just like constant flow of 90% garbage and then ten percent is the important signal, but how do you actually like really get that out of, you know. Um Yeah, no, like I said, we're just we're reaching out and just trying to get some perspectives from the language, we talked to a couple of we talked to some people in SEO and um education and so it's just cool to hear like how how people are how they're how stuff is training and how they're adopting and what they're using and things like that. So thank you very much for coming to show him. Um today and Sam RC does have any questions that we'll get under you. Yeah, I thought it was something. Yes. Yeah, it makes me think I think today made me think about uh well uh first of all it's been at 30 minutes, so if you if you had to draw then um Okay, cool. Um the other thing I was thinking about when you mentioned Tasemaker, I mean some of the things we were talking about earlier with um kind of the reboot of D work and and uh some other common so my brain starts working like so I'm getting uh my contacts everywhere, but um basically what I was thinking about was this has come up in a lot of our talks, how you know people are so really like for tastemaker or the person that's in curation. Um, you know, the the vision ultimately comes down and the creativity ultimately comes down to the human operator that's using AI to get mergence to you know all those things. And I well with the publisher model, yeah, it's it's interesting because it's like um at some point like you d when you almost need to delegate one person to be like to make that decision. And and then like have tape that they're gonna run in the right direction with it. It's just like normal um kill of management I guess. I think the there's something interesting in that because a lot of times we run around the same problem
[00:12:00] [andrej]: like a the um uh attention capacity uh like what exactly are you gonna pay your attention to if there's so much information and so what I've noticed with myself in particular is that I'm just becoming more and more selective about the conversations and the people that I'm talking to and uh and and like I I don't look at any group chats uh on telegram anymore. I also don't I prefer not to do them unless there's a specific like I was very clear with the we did um so yeah I think that has also changed at least in my perception a little bit more that's much more focused on who's the person that I'm talking to does that actually like is that actually useful conversation on maybe I'm artistic or something but it's seems to be in like a development that came along in the same kind of trajectory as uh AI has started to become very human like yeah it's not oh um yeah um I would say yeah sure um so in short and because this is all very like early and things are just kind of emerging um that I would say like just pieces are coming together um I basically wrote down deep player as like a thing that has existed and um try to take the parts of the most from it and check that for the people that the I mean obviously that we're just friends and um stuff around and the like I would say collectively elevating the new version of that uh that really very different ish. It's not the same. Um unless I'm not really answering the question anyways but um I also insured um the idea is to have a um uh machine for that is uh uh business that serves specifically design and um like talent related problems so like obviously using AI to um set up hypotheses or do some research with prototypes have some validate um certain technical interventions and maintain as a as a like that's the core as a business model uh at least um in um yeah networks right now and um I want that a can be like a network of experts that have specific skills and do different things for the own products that
[00:12:00] [duckanbro]: We just keep putting the bigger text or intakes and it just keeps going. Because now the we're just interneting. Right? And it's like someone on this has to be able to come in and just say this is the way. And um now we can point all our AIs in the same direction. Like Uh-huh. Right. Uh Yeah, like focus. So bringing back to that main press. Um yeah, that's all interesting. Peace. Do you have any comments? Or Sam. Kind of quiet sound for me, I don't know. No. I can crank up here, but it'll be really Yeah, that's really good. Uh how different people's missions compliment each other. Uh yeah, that's like definitely a the problem I see is everyone has a kind of looking like somebody uh doesn't thing or passionate about something and and the overlap. You know, take mention one question I thought this is usually astronomy. Like not only uh do we take uh do we take this fine, but also maybe a more optimistic one like Yeah. Yeah. That's awesome. Um cool, uh let's see what's on the questions, I think. Just wanted to say thanks again for helping on and uh as we get all this stuff put together and show how the files will drive. Um we've got a lot of we don't use any interviews to perform into the what we're calling the dark factory and such a content pipeline. Still got a lot of swap, but a couple pieces of of gold sometimes. But there's a lot of uh that kind of um, you know, like connecting through lines and threads and things that are happening and understanding how people are doing stuff, so we'll be pretty valuable. Well no, that's kind of thing with that. Cool, well thank you man, and enjoy the rest of the time in Berlin. Um yeah, don't don't uh don't worry too bad. Thanks again. Yeah, great chat again.
[00:15:00] [andrej]: businesses on like basically curating the network of um founders and helping them I guess um from a product sample and like set of private um and then there's uh part that's yeah and then there's like an ILI component so like uh essentially something like um residencies the um like how high agency high requiring people can come together and uh learn from each other develop and accelerate the progress and and develop themselves as as humans as well in a very short amount of time and then work and and and work together really um efficiently if there's interest obviously um yeah I like that yeah yeah I I think that's yeah it makes sense to me. I mean there's a kind of like economically makes sense fiber like in the same space um I think yeah I'm good. Um question uh I think the yeah publishing the probably uh publishing purpose is probably however uh we haven't set up kind of full pipeline for just like like another um this uh like a full on system that uh that is our intention to kind of build development public and showcase uh all the different things that people are creating and also like facilitate collaboration among them. Um in terms of the tech stack um because there's so many pieces to it I think it's hard for me to just uh like boil it down I guess as because I'm uh really doing like kinda like high quality development or like engineering set like engineering perspective. Um uh mostly like I said I'm using their code products and I have a clock there for uh sorry worker uh and the D1 and uh not really doing many complicated things. I have like a telegram ball that cloud have the world to feed information into it and just make regular posts. Um yeah that's pretty much like I feel like I'm very low tech on on this I feel like I should be more advanced. I mean I have like I don't know like an obsidian knowledge base where I store stuff and like I can talk to it. That's that seems to be super helpful um more so than making promotion.
[00:18:00] [andrej]: Yeah. I'm probably gonna put so that you can see as well, but generally um Yeah. Yeah. Yeah. What was the second part of uh pacemaker? No, I'm sorry. Um yeah, um I use it a lot in that service, like the service that we sent out to like applicants to either the sanctuary Ivan like online stuff. Um it seems interesting. Like I feel like doing the workshops like I did before with just setting people together and then like let's write down your two uh your vision the uh something and it's like it feels like inadequate anymore. I'm not sure how to call it, but it feels like people just want to move on with arrives and so uh like the survey seems much faster. And then kind of like it's the same process, it's just taking in a certain bit of knowledge that people have, putting it in a certain container, so like you know, for example yeah, that's your future vision and then uh just measuring them, I mean like reflecting it back and then I have found that that process is so pretty much takes way less time on my end and on the end of the people filling up the serv uh surveys. Um but it creates a very um interesting dynamic where uh like a long lines on it and there is like yes, this is what I would also see. So it kinda like makes sense because actually connecting people, it's not just connecting the text and the ideas, it's uh just mixing part of something like you know, it'll be a so in that sense it's like a matchmaking thing. Um in in more specific examples where you have that actual matchmaking, then it can ask people like what are you what are you looking at? What what can you offer what do you seek and then that says uh something you can just drop into cloud code and cloud code tells you like oh this person is can help this person, this person can help this person. Um when someone says I'm like one of the things that I played around with
[00:21:00] [andrej]: is actually just over time feed the database uh on personal information around that's your interests, that's your project, uh whatever you develop and and then uh just keep that in the database along with the profile and then just ask that database a certain question. For example like who would like to test my product, who would be a potential user of it. And then uh like jungle or whatever we use, um we'll be pretty good at finding the right people and why they're relevant to the thing that I'm asking them. Does that make sense? So like imagine you like something and you're just women kind of like find them because you have knowledge about what their interests are and you clearly have like a silent point for a conversation. Um so yeah I think I found that increasingly more important for me personally like whether it's like a personal life or just like communicating with other people like I find myself more often in the mental save of noticing that someone's like going adrift with like what excites them and I have to like pull them back to like to actually talk about this thing. Um and I'm also with myself like I have to like oh this is super exciting so it was for the side quests. And you got the main quest yet and so like that's definitely something useful to practice what the hell I used to be better that I'm planning myself doing it right now. I can really interesting question. One thing that I'm noticing is that like when default solid uh I hired freelancers and specifically look for like high agency that people just love to do what they are good at or are good at the thing that people are doing. Um and and that was told me like a prerequisite and the dynamic the surface that kind of makes them that was usually that I would be like doing the project, let's you know we need to work on that, let's put the best people in the right positions and then collab and then I just and then we have fun. And um and that was nice until I kind of like I I'd let go
[00:24:00] [andrej]: a lot of my personal or I let's let me put it this way I haven't recognized the responsibility that I was carrying as a as a founder myself like uh as someone with that ceiling so I was not confident in my own decisions and so I rely on other people often to to learn but then that was not the best way and so like then then then things like fizzling out anyway. Um so the foundation was like quite interesting because people were just really good at something but we would now it's more like basically everyone has their own company or like their own silo or their own like huge complex idea um and they're like bycoding things in addition to having like a s uh a pretty solid skill set and um so how are you like fast at getting it out there and like basically like getting something usable or like that's testable and visible and so the analytic here is more like what's the bigger thing that would always us and how can we use the tools that we're excited about bringing to the world to connect in a complementary way while learning from each other. So that's the dynamic is mostly like people build their own tools and these tools can be useful for them individually in their own audiences um and they have skills and and so the tools can be kind of fit together into a bigger ecosystem and the skills are obviously in learning from and with each other to progress individually but also as a as a collective um so I would say that's like the main difference and that I'm kind of observing from the the people around me and like that also kind of like gives me a lot of excitement and energy um like how um yeah just how unique these solutions are how power driven people are in the things that they're building, um how they're solving their own problems for themselves. So they're all like there's maybe like three four people who are doing like some something with social media and each of them has a different and each of them has a different solution. Each of them has a different kind of particular substrate and it's like it's really inside to see that coming kind of like out because it's so much easier to think and like actualize something and then yeah and then I see that as like oh these are different parts maybe we should put these people together in one and then we just use the best types together and then something that's um like you know collectively uh useful um for everyone as a lot of do not takes and uh
[00:27:00] [andrej]: understand them also I don't uh I I don't uh enjoy them. Um I uh so from an optimistic standpoint I think at least in in in my future I just want to see a lot of magical things and um um products and um environments that uh improve upon the existing condition of life and nature uh and make the most of what the technological innovation has to offer. I think there's some domains of um like innovation science uh or even infrastructure that uh so very kind of like underutilizing the eye that I mean I don't know if it's gonna be in the next six months in one year but um in one year it's definitely gonna be a lot of really interesting magical things that we can interact with that that we were previously think that that are impossible and we hold them in our hands and we'll do like this is impossible and I I'm very excited about um these moments of delight and wonder and how that um can make us better as individual humans and um make our um as as a collect as collectors much smarter and um and and more connected to each other. It's definitely gonna be random consequences like I think especially like us as like kind of nomadic online people or maybe even not nomadic but just like online native humans um we're gonna be really better at like connecting with each other and like have really interesting ways of doing that we shouldn't be gonna like a surface and yeah of course nice that's cool definitely I'm excited to read it very good. Let's go near to care I fell out there.
[00:34:47] [samkuhlmann]: have a question if you have a sec. Uh I feel like we've all been kind of working in this community.
[00:36:34] [takekek chat]: how about the future question <@435541241795444737>, 6 months from now...? (https://discord.com/channels/684227450204323876/873247103231610931/1516859200997687307)
[00:39:47] [samkuhlmann chat]: comminity becomes a protocol of compatible skill files (https://discord.com/channels/684227450204323876/873247103231610931/1516860013195427850)