Recall meeting (pending)

0:01Sebastian Schmid: There's. There's so many additional bots joining now. Record on

0:04Lukas Hoffmann: Yeah,

0:04Sebastian Schmid: those

0:04Lukas Hoffmann: I'm

0:04Sebastian Schmid: one.

0:05Lukas Hoffmann: a little bit experimenting about everything here.

0:09Sebastian Schmid: Are. Are you going to immediately replace Michael and David with. With bots with agents?

0:14Lukas Hoffmann: No, no,

0:15Sebastian Schmid: No.

0:16Lukas Hoffmann: no,

0:16Sebastian Schmid: Then

0:17Lukas Hoffmann: but

0:17Sebastian Schmid: it's

0:17Lukas Hoffmann: perhaps

0:17Sebastian Schmid: me. I

0:17Lukas Hoffmann: you

0:18Sebastian Schmid: will be replaced.

0:18Lukas Hoffmann: y.

0:18Sebastian Schmid: Exactly. I was going to say. Damn. You know. Okay. I should have known that earlier, but. Hey, Marco, We're almost complete. Just have to wait for Isaac to come. Let's hope he makes it.

0:37Lukas Hoffmann: Only. There are two bots here and they have different modes how to retrieve transcriptions or how to do the things. And I want to test the performance a little bit. What's fine for everyone here. Marco says no.

0:52Sebastian Schmid: Nope, didn't give consent. We left immediately.

0:56Lukas Hoffmann: I asked.

0:58Sebastian Schmid: Okay. Okay. Okay. Okay. How we've been doing, guys like David, I think you already had a call with Lucas, right?

1:06David Simeon: Yeah, I did. I did. Less than an hour ago.

1:10Lukas Hoffmann: It

1:10Sebastian Schmid: So

1:10Lukas Hoffmann: was a

1:10Sebastian Schmid: you'll.

1:10Lukas Hoffmann: very nice call, David, isn't it?

1:13David Simeon: Yeah, it was. It was a very good call.

1:15Michael Mensah: Good.

1:16Sebastian Schmid: You'll be the first. The first guy to be pumped up with tokens for. For Claude Max. Now you will have the. The superpower.

1:27David Simeon: Oh, that's good. That's good.

1:31Sebastian Schmid: All right. I'm just checking where Isaac is because I think, yeah, today we only said the daily, of course, for like 15 minutes. Because in the end, like, I mean, I think we already have a bit of an idea of like what we should be focusing on, I guess. And I mean, I think the main thing was for Isaac to get the, you know, guys, if something happened and he doesn't show up, then

1:51Lukas Hoffmann: 날씨였습니다.

1:51Sebastian Schmid: hopefully get to do this tomorrow. But it shouldn't keep us back. I think maybe to get started already. I think in terms of coordination, right. I think Lucas was kind enough to really create a few asana tickets. I think some of you guys have already either grabbed it or seen that it was something to be done. I think it still will most likely be sort of this loose structure for a while. But in terms of to dos, maybe we can just go over it. I think maybe I'll start because I think my main thing is that I need to coordinate together also with Tom from Tikfinity and together with Lucas to create the first, let's say, release structures. I think that would be in terms of an expectation, of course, that doesn't exist on an island because we need to coordinate again with Marco and you guys about, okay, what is a possible feature set that maybe does make sense in compact version of it? And I think once that is defined, then we can decide on how to push it into Tiffinity. But I think that would be one thing. The other compartments are a bit. More on the, on the longer term strategy is like how to kind of like push the release all the way, I would say, from clipping into this AI marketplace. But that's something that will be slowly added over time. Other than that, on my end, I just need to make sure that we get the right documents from the lawyers. Also, we have a draft for this already, but not a hey, yo. Hey, Isaac, nice to meet you. Well,

3:30Marco Cimolai: 80.

3:30Sebastian Schmid: unfortunately, Isaac, now, because you're the, the new guy and, and the route, we'll have to start with a little bit of an intro because most of us, I think, know you, but not everyone maybe, Isaac, you want to say like two sentences about who you are,

3:47Isaac Kogan: Well, this is going to be awkward.

3:50Sebastian Schmid: We can make it as least awkward as possible. I don't know how we do that, but

3:54Isaac Kogan: No, no,

3:54Sebastian Schmid: it.

3:55Isaac Kogan: no. The microphone wasn't working properly. Okay. There we go. Hi. How are you guys? I'm Isaac. I've been working with Sebastian and Lucas for a little while and before that, David from David. And yeah, I work on a lot of reverse engineering. I work on front end and back end development and I'm pretty excited to be here today. That sounded like aa.

4:19Sebastian Schmid: That's all right. That's all right. We'll take it.

4:21Lukas Hoffmann: And you built a whole infrastructure around recording the VODs and all the magic here. That's the reason.

4:28Sebastian Schmid: Yeah. I wouldn't want to. I wouldn't want to glaze Isaac even further, but I think, unfortunately, like, similar to Marco, like, those two are the main integral parts of this equation for now, because they're sort of the. Whatever we're building today is like the brainchild of those. Like Isaac, as I said, comes from the reverse engineering side, and he's like the main capability to make sure that we even have something to look at and to potentially work with. And, of course, Marco's role, you know, but maybe Michael or David, you want to go next?

5:02David Simeon: Okay.

5:02Michael Mensah: Yeah.

5:02David Simeon: So. Okay. So should I go,

5:07Sebastian Schmid: Sure.

5:07David Simeon: Michael?

5:08Michael Mensah: Yeah,

5:08David Simeon: Okay.

5:09Michael Mensah: you can

5:09David Simeon: Yeah.

5:09Michael Mensah: go. Yeah.

5:11David Simeon: Okay. Good afternoon, everyone. My name is David. I'll be on the project as a mobile developer with Flutter. It's nice to make acquaintance, Isaac.

5:24Michael Mensah: Yes, I think I can go. Next. So my name is Michael Mensah, so I'm also from Ghana. And I'll be, you know, part of the product team. That's me taking part of the, you know, product design role. And Nice to meet you. ID too.

5:39Sebastian Schmid: Cool, cool, cool. Yeah. I mean, maybe Isaac also a little bit of a recap for you. Obviously we're a small team, we're just getting started. So most of this is basically in motion and in flux and it's not so strict anyway. But in terms of our responsibilities, I mean, you understand and you have seen the recorder project. I think what we have now decided is to really kind of build it basically like bottom up in terms of. Okay, we'll start with clipping as the first, let's say layer because it's the closest to, I would say, what the Tikfinity users need today. So it gives us like a good edge in order not just to quickly increase revenue, but also to bring in new users to get consent to recorder. Right. So they would say, okay, I've been a streamer on Tikfinity, now I can do it, I can actually record store and clip my content. So therefore it's like a good bridge over into this recorder world. And since Marco, I mean you guys have met each other, but, but he's the natural lead in order to make sure that we kind of get the right features and so on and so motion. Then eventually what we'll do is from a strategic mode, once clipping is set up, I think we want to get sure that we have a low level integration with tigrinity and then we'll have a mobile and a web app of recorder where sort of more functionality is provided. Then we'll add sort of the whole content moderation, call it Stream, Recorder, live Recorder, whatever, use case right on top of it where we'll bring in additional users to kind of like interact with that content. And then at the end will come this AI marketplace use case which for now at least on the MVP version, Lucas builds on. Right. So he is basically on his own island, like starting to make sure that we have a POC that we can test and play around with. However, of course, which I mean, no secret, we all need to rely on your infrastructure in order to even have something to look at, even have something to record and then eventually push it into, into bigger production. And then, yeah, like Marco will be the lead for the, for the clipping and also for the, for the web app. Michael is UX UI designer who will help us with everything around corporate identity, screen design, etc. Etc. He's really the, I wouldn't say the odd man out, but unfortunately the guy who is like blocked the most because we have really like, we hired an agency and the agency is like not done, which is basically our fault. It's not their fault because we haven't settled fully on a corporate design. And then David is responsible for everything around mobile. Yeah. And I think that's as a role distribution. Which also probably explains sort of the. Because I don't know if you've already seen Isaac, but we have set it like an Asana board. Again, as we said yesterday, this is not a top down dogmatic. Oh, here's a ticket. Do that. Da, da, da. It's more like, okay, these are like pillars to just like have some orientation about. Okay. Most of the people are pretty much. Everybody knows what they would be working on anyway. And I think there'll be more collaboration over the next coming weeks. Of course, right now, I think we're basically stuck a little bit in this. Prep, prep phase or, or I don't know, trying to kind of like get our, get our setup up and running. Yeah, so far so good. I think other than that what I would suggest is we can, I mean either we do it like super, super awkwardly go about it like one on one is like oh what am I working on? But I don't think that's necessary. I think what we should be doing is like really just elaborate or think about what the next steps would be. And I'm aware at least of Lucas and David again having had this call earlier about like getting some of the infrastructure and maybe like yeah, architecture discussions going already. But I think you Isaac ideally should coordinate a little bit with Marco and with Lucas I would think eventually also with David of course. But like in terms of like what APIs and I think maybe, maybe Lucas, you want to say a few words?

9:39Lukas Hoffmann: I mean, there is a document I already shared with you. Isaac and Marco gave some comments into that as well. And I think at least a lot of these comments are right. That means perhaps there is a small discussion needed between Isaac and Marco to discuss

9:57Marco Cimolai: That.

9:57Lukas Hoffmann: or finalize on that. Because for me the comments are fine. And yeah, this API endpoints would help us to really use it. Like okay, we can start a recording. There is perhaps the content and compliance management. That's something perhaps we need to discuss if it should be on your layer or is it like more in our backend API layer? I think it would be helpful to have it on your layer because if we want to, I don't know, extract data for labs, it would be really nice that we already have some information if you have the concerns and don't need to bubble it down to the API. But yeah, I

10:33Isaac Kogan: Yeah,

10:33Lukas Hoffmann: think.

10:34Isaac Kogan: I see that comment right now. That was basically the thing I was going to flag is that we already have a mini consent layer built into it, so depending on what we decide has to be removed. I half agree with Marco that dogmatically it makes sense to make it a collection only layer, but we may want to do that at the end since we need AI labs. That may be something that we change in the future for purity's sake, but for now we kind of need it if we're going to be doing both at the same time because we just don't have the API built out yet. So we would need something like that within this exact one so that you can proceed on your part while the recorder is being built.

11:17Lukas Hoffmann: Yes, and for example, the storage. I mean theory. We could do the storage management as well. But for me it is more like the pipeline needs to decide if I can record or if the storage is full. Perhaps it makes sense to have this storage controlling layer in your software as well and not on the API on our side. But I think there are some things that are not really needed now for the first moment, like the storage management. That's more like the later requirement. I think what really would be nice would be having API to mark a streamer as you should be recorded and fetch the information about past bods and all the how to retrieve the information that we can use it in the app that we can use it in other situations. That would be. And perhaps. I'm not really sure Marco, but do you need else. Is there something to make the cloud. Cloud connection or to check the. The transcripts if they need. We need to think about it. That's. I think Marco and Isaac needs to coordinate which endpoints. Marco needs to work on the clipping infrastructure.

12:32Marco Cimolai: Yes. There are a lot of things to discuss. I don't know if you can do it now or.

12:39Lukas Hoffmann: I think it is more. It makes sense to make it a smaller round because yeah, there are not everybody should be involved. I'm not sure if Isaac has the time because we said it should be a smaller time, smaller slot now and then perhaps we need to coordinate afterwards Marco, Isaac and me to make another call to discuss that.

12:58Isaac Kogan: Yeah, that could work. I would say that most of these endpoints are very small to make like we're talking like five minutes per. Just because none of them add complexity. They're all just existing views of data that's already stored in tables. Nothing really modifies anything in any major impressive way. So it shouldn't be a problem. But we can call afterwards about it. The other thing I want to flag while we're here is this may not be worth it now, but in the future, you, you are definitely going to find that storage costs are going to be the main cost for everything.

13:31David Simeon: 19.

13:31Isaac Kogan: And one, one way that I know of handling this is if we were to set up our own storage servers. You can rent fairly low powered servers that have a lot of attached hard drive storage, which could be a solution here, because you can install S3 onto them and turn them into S3 servers, which is the standard way of doing this at mega scale. We're not at mega scale right now. But when the bill starts hitting, you know, $30,000 a month, you can easily turn that into $10,000 a month by switching storage. But then it's too late. So that's, I mean, I, I don't know how to make that business

14:11Lukas Hoffmann: I

14:11Isaac Kogan: consistent.

14:11Lukas Hoffmann: mean, it's never too late, but it's more effort to switch.

14:14Isaac Kogan: Right? Switching $30,000 worth of storage is switching like, I don't know, a couple thousand terabytes. So it's, it's a little hard.

14:27Sebastian Schmid: Yep.

14:28Marco Cimolai: It also depends on what we need to do with that, like, if it's infrequent access storage. Even Cloudflare has much lower pricing for infrequent access storage. But if we need to, like, serve this to viewers,

14:43Lukas Hoffmann: I mean, that's, that's the discussion that Sebastian I we had already 20 times because we we have some use cases in place for AI labs. We don't need the data really often to access them. But if you want to have the VOD case and people should watch the VODs, of course, you need them hot somewhere. But it is not totally clear yet how big the storage will be per user. I mean, they're paying users as well. And it's like the business model is a little bit hard to understand for us right now. That means there is no final discussion yet. But I think for clipping, for example, you don't need to store everything three months, because if you clip a stream after some days, not months in the past, that means we need to think about it, how we go with that. That's the person you want to say eventually,

15:34Sebastian Schmid: No, no.

15:35Lukas Hoffmann: but.

15:36Sebastian Schmid: I just wanted to add a bit without making it more complicated, of course. But I think the idea will be, or at least I think what we came down to is almost like a triage system in a way, right? Where we say, okay, we have maybe like a certain type of content that we'll always keep for longer than there's a certain type of content which will probably delete after a week or a couple of days, and. And then we flag content that maybe is around. And then ideally, again, a little bit dependent on the business model itself, right? I mean, like, I think maybe David, you might already have seen this. Isaac, I know, knows this, and Lucas as well. But with somebody like Stream Recorder, right, I think the way that they. They monetize it basically is like they have the viewers pay for the recording. So it's like, okay, I want to keep access or I want to store streams of a certain viewer of a streamer. And then of course, they essentially pay for the storage cost, right? So this is like a somewhat of a nice way of like sort of transferring the cost over to the audience. But I'm not 100% sure we can do that or want to do that, especially not from day number one, because from day number one, we'll have creator focused only. And even though we might not differentiate between like, in terms of you're either a creator or you only an audience member, because ideally you're both. But in terms of a business model, I think we'll have a very straightforward creative clipping only part. Now, what could make it slightly easier in terms of the EI case is that some of the partners we talk to, which are mainly data brokers themselves, they offer to actually take the load off of us and actually store it for themselves. Like,

17:10Lukas Hoffmann: Yeah,

17:10Sebastian Schmid: no

17:11Lukas Hoffmann: but

17:11Sebastian Schmid: pun

17:11Lukas Hoffmann: I

17:11Sebastian Schmid: intended.

17:11Lukas Hoffmann: don't think

17:11Sebastian Schmid: But

17:11Lukas Hoffmann: that they know how

17:12Sebastian Schmid: it's

17:12Lukas Hoffmann: many data

17:12Sebastian Schmid: like,

17:13Lukas Hoffmann: we have. Because normal other players they don't have perhaps.

17:18Sebastian Schmid: That's

17:18Lukas Hoffmann: Peterbilt.

17:18Sebastian Schmid: true. It depends on the data set size, right? I mean,

17:20Lukas Hoffmann: Yes,

17:21Sebastian Schmid: think like, for example, now

17:22Lukas Hoffmann: but

17:22Sebastian Schmid: we.

17:22Lukas Hoffmann: yeah,

17:22Sebastian Schmid: Yeah, but anyway, like,

17:24Lukas Hoffmann: we.

17:24Sebastian Schmid: that's

17:24Lukas Hoffmann: We

17:24Sebastian Schmid: like.

17:24Lukas Hoffmann: will think about the best structuring of the costs. And I mean, if you're a really good working project, 30k of storage cost is doable. But yeah, if you can. Cheaper, of course we should do that. That means Isaac, perhaps. Is it the. The same company? What you mentioned with the. I think you sent it through already with the solid service. Okay.

17:44Isaac Kogan: Yeah, I think I did.

17:46Lukas Hoffmann: It's only about orchestrating its infrastructure. You need to have someone who is doing that. It's not like that. And you have a S3.

17:55Isaac Kogan: You traded off for, I guess, the hosting costs of like the the developer that's responsible or the DevOps engineer that's responsible for managing it. But I think that when it comes to like, storage servers specifically, it's battle tested hardware, you're most of the time when you have an issue, it'll be an issue that the host has to solve, not the DevOps engineer. Just because this S3 layer is like something that is installed millions of times per day by CICD pipelines onto servers and automatically updated. So I don't think that it's going to be the source of failure here. I think more likely the storage or the network, or any other number of things that would have to be diagnosed. But anywho, yeah, we'll figure it out as it comes.

18:40Sebastian Schmid: Okay. I mean, I think other than that, I think all. I mean, I'd be happy if each and everyone at least has a rough idea about, like, what they should potentially look at. And it is not lost. I think that's already a big win. I think, as a team, the rest will figure out. I think, again, it requires a lot of coordination, but I think, I mean, I'm just looking at the round, but maybe you guys can tell me. But if it is somewhat clear what to look at, then I think we can also conclude for now, because this is really just an intro call and getting to know call. And the rest we can coordinate, like, either bilaterally or in slack or on asana or whenever something's required.

19:19Isaac Kogan: Yeah, so that's it for me. But I just want to mention to Lucas I think we should have a really short talk after, but that's about it. Yeah.

19:28Lukas Hoffmann: I cannot talk directly after, but a little bit later

19:34Isaac Kogan: Yeah, sure.

19:35Lukas Hoffmann: we can chat. I

19:37Isaac Kogan: Yeah.

19:37Lukas Hoffmann: will send you a message.

19:39Isaac Kogan: It's just. It's about TikTok and connection stuff, so that's all.

19:45Sebastian Schmid: Good.

19:45Isaac Kogan: Okay.

19:46Sebastian Schmid: Cool. Well, then, everyone, thank you. And, yeah, let's reconvene soon.

19:52Lukas Hoffmann: Thanks. Bye bye.

19:54Sebastian Schmid: Bye. Bye. Thank you, guys.