---
visibility: public
---

Speaker 2  0:01  
Every every team meeting updates an LLM wiki for your company that's like the most basic piece of this product, okay? And the

Speaker 2  0:23  
good parts of this are, if I like brainstorm an idea or I commit to an action item, it's both tracked inside, like the notes for that meeting, and it is tracked in the like larger like task list for the team and company, where we're keeping we're keeping all that up to date, okay? And another aspect is, if I share an idea, it adds it to like the company idea base, and it's also in, like the notes for that meeting, if we ever want to go back to the archive of the of the notes from for that meeting. And additionally,

Speaker 2  1:20  
if I am, like, disagreeing with a teammate about something, or we have, like a conflict about something, we we track this also in sort of the LLM wiki for the whole company, okay.

Speaker 2  1:40  
And so we sort of have a Wikipedia page on like the long standing relationship gripes and the long standing vision documents. And every time someone adds to the vision it updates the big vision document. Every time a pattern of a particular challenge or issue occurs, it gets documented in the wiki, like if someone doesn't respond for three days, or something like that. And this annoys me, I'll say, Yo, you didn't respond for three days. And I would say this is an instance of the existing gripe that I have with you, or whatever. Does that make sense? So sort of, is this a creative knowledge base for your company that sort of versus friction? Well, friction is just a little piece of it, it's, it's more. It's more than that. It's, it's, it's a sort of persistent updating knowledge base for your for your team that is well structured and and, and updates things. Now there's a lot of ways that you could, you could do this,

Speaker 2  3:13  
I think, really nailing the workflow where you go from wherever you're recording your meeting notes to updating this is is is like a question that we need to figure out, because, like, if you use Gemini meeting notes, it all ends up in a Google Doc. If you use granola, it ends up in a different place. If you use otter, it ends up in a different place. And so there's, like, some questions to be thought out around there. Now I'm trying to distinguish between this, which I could just get the entire team to, like hyper focus on that, which might be the right thing, and I think it might be worth doing, okay, or we can continue with sort of parallel tracks. And so the possible parallel tracks are this.

Speaker 2  4:21  
They're all adjacent to what we just described. So we have a Slack bot for the team, which I think is essential, because it can reference the LLM wiki for the team. Okay? I and it actually surfaces the information. So that's just like table stakes, and it can also include all of our history that's in written format as well as the meeting format.

Unknown Speaker  5:00  
I um.

Speaker 2  5:07  
And then the other aspects are this, we've been thinking of making a version that you can put an iMessage on a group text, okay? And there's a lot of applications of it, but it's basically like you just need a bot that you can add to an arbitrary group text, and it will be able to maintain the knowledge from that group chat. It could be text, it could be WhatsApp, it could be whatever else you'd like you

Unknown Speaker  5:43  
Additionally,

Unknown Speaker  6:00  
you

Speaker 2  6:07  
can. And I think, I think, by the way, to get around some WhatsApp Terms of Service publicly, we could advertise it as a human assistant service, that there's an assistant that you add to the group chat, right? But that happens to be assisted with AI, because otherwise WhatsApp bans fully bought users, I think, isn't that kind of interesting? Max? What's that? Yeah,

Unknown Speaker  6:39  
I feel like, I feel like WhatsApp is useful for for a personal agent, it's, it's not so much useful,

Speaker 2  6:50  
right? And so and so I'm trying to think about, like, I think there's a million use cases this thing. And so I

Unknown Speaker  7:04  
yeah, having,

Speaker 2  7:13  
I guess Are you part of any WhatsApp groups that are really high signal? I

Unknown Speaker  7:23  
some WhatsApp groups, yeah, like WhatsApp groups that have a lot of knowledge in them,

Speaker 2  7:32  
retrieving for for like I let deep research agents verify and solidified. So so the other possibility, by the way, is with WhatsApp groups or whatever, you could just do open class style, where it just takes your own WhatsApp output and no bot actually needs to sit in the group. And so that's the other option. Is, like, you just have it be an open claw that is on your end, and it's, it's sitting there, and it's maintaining the knowledge as you, as the human user. It just you have the bot on your end. So that's the other possibility, which is, that's a, that's a larger idea here,

Unknown Speaker  8:30  
the Oh man, I'm just thinking.

Speaker 2  8:42  
So. I'm trying to sort of prioritize between so I think iMessage is very important, at least in the US, and having a bot that you can add to a context that is not your personal bot, and it sort of has. It has the ability to

Speaker 2  9:18  
not share all of your personal information, but to share some of your information and ask for your approval about sharing information. So if you're in a group text and you're trying to schedule something or a group email, you might want to share some of your calendar, but not all of your calendar.

Speaker 2  9:50  
And so the bot would ask for approval about, can you share your availability on this date to help schedule. And then it texts you, and you just click Approve. Okay, so having sort of a cross privacy scoped bot is a piece of the product that's there's a lot of different use cases and user stories here for very similar, overlapping things. And I think it's very helpful to be looking with a lot of precision about which use case we're talking about. Is it a B to B? Use case, is it a B to C? Use case, is it in the B to B? Use case details like, do you want to integrate with Google meet Do you want to integrate with with Zoom meeting notes? How is it going to get your Zoom meeting notes? Do you want to build our own meeting note taker? Similarly, with the bot, it's like, are we focused on WhatsApp chats? Are we focused on iMessage chats, signal or whatever? I mean, they all are important, but we got to start with one. And the question is, is like, should the bot be something that sort of, it's like your open claw, and it automates your own I message that's like option one, or your own deeper bridge, and it automates Your own, like WhatsApp history and or do you do you actually want to make a number that you can add to this arbitrary group chat? And so I think there's a benefit. There's maybe a place for both of these, but these are all places where we can think about cutting scope and focusing to get something that's really good to market. If we can be a little strategic here, so it'll save us some time to plan it. Okay, so on the service, on the certain the next step of planning. And I'm just really rambling here, and I'm giving you my full overview, there's one other important thing, which is, was two other important things.

Speaker 2  12:10  
Well, two of these are details, so they're not as important. The details are. Do we do the wiki hub structure versus the nuos structure, which is more of a graph, that's a detail we can discuss that later, and then the other the other detail is,

Speaker 2  12:34  
Let's graph the the wiki hub. Just Just Just bear with wait for a moment while I'm thinking, yeah, do we have a I've been building this this real time meeting Outliner, which is a custom user experience. And it transcribes your meeting, and it takes structured notes on the meeting, and it updates your knowledge base, which you could say your updates your LLM wiki for your team with every meeting, which is kind of nice. Do we need our own custom note taking app for this? Or should we just focus on integrating with whatever existing note taking apps people have? That's a strategic question. And then lastly, the sort of real time assistance in the meeting. How important is that like having an ambient information display that pops up stuff on screen and reveals relevant information in real time, or does research fact checks and build stuff? How important is that as part of the user story versus just tracking commitments. I suspect that with all of this, it might come down to some 8020s in the actual market where one of these works the book works first. But it's good to think of the different possibilities and figure out which are most essential. Okay, for the sake of complete comprehensiveness.

Speaker 2  14:30  
And by the way, I'll give you a dashboard showing the outline of all these. In just a moment, I'll give you a page showing the outline of all these. And for the sake of design of our software, that would be a great use case of our software, as I could say, Hey, give me a dashboard of all the key branches that we could be going down, all the key possibilities being considered. That's an example of a real time prompt that you can put onto this sort of AI ambient display dashboard, which that could be its own product. Forget meeting note taking, just an AI ambient display dashboard that keeps track of things, all right, so the next one is and by the way, I could just say, hey, AI go build that right now. If I had my system thought out in the full the fully fledged version of the system, I could just say, hey, go build that node. Go build that whole product for us, which is kind of cool. The last thing that I haven't talked about is our mobile app, which is very similar to the app that we currently have on the app store called thought stream. Our new app is called note stream, and all it is is it's a collaboration first and LLM friendly version of our old app, thought stream, and so I can be taking my notes. It's almost like Slack for myself, tag stuff, filter stuff. It's really just slack for myself again. And then I can spawn an arbitrary agent to go build something or flesh out an idea in the thread next to every one of my notes. And I think that's pretty, pretty useful, having something that's like Slack for myself as a different agent deployment dashboard. And oh, right. And then there's one more thing, which is what I call better GPT. And it's just chat GPT, but it's cloud code under the hood, meaning it's a coding agent or Codex under the hood, and it's running on a private server, and so it can do all kinds of normal stuff that a coding agent can do with normal GPT can't do. And I can build a connector to normal GPT and normal cloud so that it can do coding agent stuff. It's just like a cloud box for your GPT, that's like an entirely other product direction, that every one of these is valuable, and every one of these is synergistic, and a lot of these share infrastructure. So that's the full that's the full outline so far.

Unknown Speaker  17:17  
So for the channel, do you want to try Mr. Lowest from your WhatsApp. How do I do that? I.

Transcribed by https://otter.ai