Episode 1: How we built Kimiko

1:23 Who we are

1:23 Cuong: I'm Cuong.

1:25 Max: And I'm Max.

1:26 Cuong: We built Kimiko. We built Kimiko. So the whole conversation today is to walk you through how we built it, the whole process, who we built it for, how we built it, the challenges and the wins, and also the roadmap, and also what we are most excited about.

1:43 Why we built Kimiko

1:43 Max: Well, the original idea was we do a lot of meetings, and my sister also does a lot of meetings. We are all sitting at home in Ukraine, and we use a lot of tools. We know about Granola, and we have to-do task management and stuff like that. And so we're playing around with the idea of how to make your life easier, especially that I have a bit low attention when it's a very long meeting, two or three hours. And we wanted an app which would replace four apps kind of thing, because we pay for many other ones. Yeah, that's why the idea was born, and we wanted something to be an assistant in the background, a sort of whole second brain for all your calls, meetings, maybe something you're learning, and AI to grab everything and then tell you what you need to do kind of thing. And I remind you tomorrow, so you don't forget what was your meeting, what did you guys discuss, what decisions you made, sort of all in one tool. There are some on the market which are already present, like Granola, a very cool tool. We must give them credit to that, but we saw some features which were missing, and that's why we built it. We also didn't like that everything is going to the cloud. Some, I think, calls need to stay private. Originally, Kimiko was a completely different product, to be honest. When we originally discussed, it was more focused on the aggregate AI tool or folder for many project management, but we pivoted.

3:21 Cuong: That's good. And yeah, tell us more about, you know, what are the challenges when we built it?

3:27 First challenges

3:28 Max: The biggest challenge, well, actually with you, which we had was, because the original idea was yours, when you did Scribbler, I think. And we were playing around with voice recognition and transcription in different languages and stuff like that, which we had some performance issues on mobile. Remember that?

3:48 Cuong: Yeah, it was an iPhone app.

3:50 Max: Yeah, an iPhone app. So our biggest initial challenge, whether it's actually possible to get proper voice transcription, because we know we could handle the AI part, either locally or with cloud. So the initial one was studying which ASR model, audio speech recognition model was good and appropriate and can run on device. Yeah, I think that was the first challenge and we were really excited when, actually, we saw it being accurate. Accurate, so grabbing 90% probably of the meeting. And the next challenge was building for both platforms, I think, and picking the right stack. So we also, we tried to do native for iOS and then for Windows, but we came to a conclusion that it's a bit too much to manage two separate projects and maintain core business function. So that's why we picked Tauri and everything is in Rust for performance reasons. The app is quite small, actually, the way our models are big when you start working, but the app is quite lean and performance. That was also something we spent a lot of focus on.

4:59 The wins

4:59 Cuong: Of course, I mean, now the app is polished, the app is published to the public, you know. What were the biggest wins or the most satisfying part of this whole process?

5:14 Max: Satisfying process was actually releasing it, probably. I spent quite a bit of nights making it and when it comes together and spent a lot of time squashing bugs and inconsistencies between Mac and Windows. So I think the biggest win is actually for us to get it in the client's hands, finally, and having people excited about it as we are, we are biased. But when we see other people excited about it, that's a cool, cool feeling.

5:43 Cuong: The biggest win to me is because, you know, building an app is one thing, but figuring out the architecture, the what's next, you know, like the how we build it for enterprises use enterprise grade. Right. I felt that that was the biggest achievement because anyone today can build an app with using AI, right, but figuring out these parts, you know, building for the bigger purpose. The bigger purpose or the bigger use of enterprise or, you know, public users and

6:20 Max: this is the key really to me. So it's exciting when you also hear from your clients or even your internal team that this will be a cool feature and it's actually built something that solves a problem. So, uh, uh, yeah, it's definitely something that the team effort here.

6:38 More than a meeting app

6:39 Max: Well, yeah, I think also the key, uh, feature, which you added those, uh, I mean, we originally wanted to be specifically for meetings and, uh, we, uh, we discussed with you on pivoted to be a note taker in general. So you don't, it's not only for voice recordings can be like a replacement to notion or to obsidian where you can store everything there. You have one app. You can, and you have dictations. If you want to use voice, you can, uh, you call it, have all your research there. And, uh, yeah, we wanted to be as a second brain for pretty much everything. That's so valued to you. Yeah. I think that was a good change and yeah, we expanded a lot on note taking functionality there. Um, they can even use, uh, like late latex formatting. If you are into research and math and stuff like that. So, um, yeah, what else?

7:29 The hardest parts

7:32 Max: To share about our development process. We hate windows. We will tell you that.

7:38 Cuong: Uh, no, I think it's a, it's the biggest challenge for developers to develop cross platforms. The advantages of Mac is that it's a unified, right? So, uh, the devices, their, their versions is, uh, is, uh, yeah. You have one manufacturer for the devices as opposed to windows and all, you have multiple versions.

7:57 Max: And yeah, it's a consistent hardware, at least because we have a good, uh, so love and hate relationship. And with those, if you have a good computer, the app is work, but we had a lot of problems with, uh, the new laptops where we have using a bit more resources than we wanted to happen. And it's not just in, uh, our problem. It's, uh, a lot of competitors have the same problem windows. So, but yeah, it does have, uh, cause us a lot of headaches to maintain.

8:28 Cuong: So I, I assume that this is the, uh, one of the biggest challenges that we had.

8:36 Max: Well, yes, probably, probably one of the biggest for now.

8:40 Cuong: One of the biggest challenge, um, maybe not on, I mean, it is on a technical perspective, but, um, can you, if you don't mind sharing a little bit of, uh, your personal life, how it is to, to be, uh, how it is in, in Ukraine right now? While you're building Kimiko as well.

9:00 Max: It's a bit stressful for everyone. And, uh, the sin for me to complain because I am in the center of Kyiv, which is, uh, not the worst place to be. There's a lot of strikes and it's stressful. So consciously we get to do it after three years, of course, but we have all the strikes are during the night sometimes. So to your sleep schedule is a bit messed up, but I usually tend to try to work during the night, uh, when Kimiko distract myself from the outside reality. Uh, but, uh, yeah, also from the privacy conscious part. Now we have like a lot of, uh, hacks attacks on, uh, the government infrastructure here and stuff like that. So it was also one of the ideas behind it while keeping it local. But yeah, we'll try not to dwell on the, on the settings and, uh, continue working and hope for a better future.

9:48 Cuong: No, but you, you know what I think, uh, already, but I said it again. I think it's very admirable, uh, when you're doing.

9:54 Who we built it for

9:55 Max: But Cuong, share your, uh, background on why you thought this idea was cool and where it came from, because we were thinking about the same thing for quite some time, but from different perspectives.

10:09 Cuong: Yeah. Um, so initially I, uh, I tried to, uh, to build script. So the whole idea was to have something on device. Um, and the, this app is intended for local use to solve some of the issues. And so some of the gaps that we saw today on the market is that everything that you use on the clouds, um, is being processed, is being used to train AI. And, um, when I was, uh, working at, uh, uh, Incorporate, some of the companies, they're very reluctant to use AI because, because of that principle is that the data they have is very private, very confidential. So it contains data of clients, of, uh, users, of, uh, patients. So when you use AI, all those informations can, you know, be leaked and, and can go to, to, to cloud services and that makes, um, the data unsafe at some point. And I realized, you know, that there is, uh, a market out there that we can, uh, fulfill. we can build solutions for that not only preserves the data safety, but also data sovereignty, because they own companies and users own this data, not the AI service providers that they're using. So the whole idea was to build apps, an app that basically combines functionalities or tools that allow users to be productive, either you are a professional working for yourself or you work in a corporate, but also it's private enough without going to the clouds, without licking them, and everything is processed on device. So that was the whole idea when we started talking about this. So how we can make a product that answers those questions and the market is there, the need is there. Yeah, we build this for that.

12:16 Max: I agree with you. And we talked about a lot of clients that share the same sort of opinion on this, that they don't want to leak misinformation. And yeah, the issue today is people giving a lot of trust to AI. They're integrating into every tool chain, and you'd have no full picture of what it's doing. I mean, sometimes you give full access to the operating system and pull any information. And by being completely local, we remove that issue completely. Your data is much more safe by using it, but we think there is definitely a use case for an app like this, for the private sector again, and for government as well, private companies who have IP to protect. Yeah, that was the idea behind it.

13:12 What's next

13:12 Cuong: You can touch base on also on our pipelines, what we're planning for in the future as well with this app.

13:19 Max: Because regarding our roadmap for the next features, I think we're going to be revisiting the mobile app another round. Probably initially we'll target the iOS. We'll see, maybe Android will be also on the same schedule. But yeah, mobile app, which will...

13:36 Cuong: How about Windows Phone?

13:39 Max: They still exist. Hopefully not. I don't want to buy a Windows Phone. For the users still on the Windows Phone, please switch to something normal. So it's a face-to-face meeting, you can record it, and you have all your notes as well. It's a gap we currently have, and we'll be addressing it, I think, in a month or two, to release an app there. And also, we're discussing local a lot and privacy. But we understand the gap that not everyone has the compute. So we're looking into a very private and transparent cloud providing service. With a full audit, so people know that it's safe to use. I think we will be working, trying to build our reputation on being, what do you call it, taking care of your data responsibly. So yeah, I think we will be doing a bit of cloud integration, as well as expanding on the collaboration features within the app. So people can share docs, edit docs at the same time, assign tasks, and keep track of them. And also, I think we'll be adding a bit more, still under question, but I think we might be adding some more integrations with other cloud platforms. Especially for enterprise, if they have some pipeline workflow that they do, so that our app plays nicely in their current existing workflow. But yeah, that's what's currently on the roadmap, I think.

15:13 Cuong: Yeah. And we also talked recently about integrating some additional tools, you know, some additional features, for example, a teleprompter, or even a functionality that helps in interviews, job interviews, for example. Right?

15:27 Max: Oh, yeah, yeah, I forgot to mention. Yeah, sometimes you're in a meeting and you don't know an answer. So, and since we have a full transcript, we want to add some quick shortcut to save your butt, or to help you research something on the side while you are working, in the context of the meeting. So, of course, this will be an additional feature we're looking into, because every business has their own use case or their own needs. And I think we'll be making our app a bit modular to cater some functionality to a particular client. So, we're flexible.

16:02 Why “Kimiko”?

16:02 Max: Well, we named it after our cat, whose name is Kimi, who keeps us sane in the dark days. And he's named after Kimi Räikkönen, which I was a big fan, because initially I was pursuing a racing career in motorsport. And I was sort of a sport idol for me back when I was a kid. So, well, not a kid, a bit younger than I am right now. Yeah. So, we named him after that. And he's very character-wise same as Kimi Räikkönen. Very...

16:37 Cuong: So, yeah.

16:42 Max: We named him like that. And I couldn't... When I started this project in the show, created the repo, I couldn't come up with a name. So, I named it Kimi Räikkönen. And I shared it with you. You said it's cool. And... Yeah. That's pretty much the story behind it.

16:56 Cuong: And how about the logo, dude?

17:01 Max: I was about to say that Kimi Räikkönen has a nice ring to it. So, that's also a thing why we picked it. It's easy to pronounce in many languages. So, it's quite a good name. Regarding the logo, well, we wanted to make it look like a cat. And we tried like a hundred iterations of them or not to do what we wanted. So, in the end, we went very basic. We just took a circle, split it up and rotated the parts. That's pretty much it.

17:33 Cuong: No, it does, honestly. And hats off, you know, because neither you, neither me, are designers. Graphic designers, okay?

17:45 Max: Look at the logo.

17:46 Cuong: There's four little parts, right? So, that is the representations of four tools that we integrate in one. But we only discovered now because we're talking about it. But it was not intended.

17:59 Max: Well, yes, we do have our app replaces four different apps in the functionality, right? So, the four pillars of productivity.

18:09 How we're different

18:09 Cuong: You know, I got asked the questions like, how are we different from competition? Look, I mean, we are not that different, but we are different in a way. Because, first of all, I mean, what differentiates us is that everything is running local as opposed to the others. They all run on cloud. And you have no control on your data. You know, the moment you stop the subscriptions, what happens to your data? And also, you're depending on the internet as well. Of course, I mean, you don't do meetings today, you know, online meetings without internet. But all your notes, all the things that you build throughout using our app is available to you anytime. You just open your laptop with or without internet, you get that. Like, that's where we differentiate ourselves. Our mission right now is to build sovereign AI solutions for users and enterprises. And this falls into that, you know, that mission.

19:10 Max: Yeah. What you mentioned that, yeah, you have no internet. Naturally, you're not going to be on a call. We have no internet. But you have access. And you also have access to that dictation. And like we are in Ukraine, sometimes you have no power and you have no internet. You can still continue working on your laptop. And yeah, you're a bit limited, but you still can access, you can still query, you can still write your doc and continue working. And that, I think, is a big difference between them. Yeah, we're losing a bit on, you know, milliseconds on processing time. But everything is local. No matter where you are, you can still do it and still work. So, yeah, we're a privacy-focused AI note-taker mixture of...

19:55 What excites us

19:55 Cuong: The moment when we finished it, or even before, when we got the integration of all those features, I got so excited. I hope that, you know, people, when they use it, they see the benefits of it and they can make it the only app that they're going to use.

20:14 Max: I agree, I agree, and I hope it's never final, so we'll keep improving it. Man, it's a goal. You should never stop improving a product. And yeah, we hope you guys like the product.

20:29 Cuong: How about this, Max? Next time we record this again, we show some of those functionalities and we demonstrate them so everybody sees what it is capable of.

20:41 Max: Yeah, we should do a founder demo, I think, and let's show off some of those things. And because, honestly, I don't remember all of them already. And we have a lot of other people on the team, which also provided a lot of input to help us address some critical. Hope it's something worthy of your time. Excited time. Subscribe to our YouTube channel.

21:06 Cuong: That will be the professional advice. Becoming founders. Becoming founders.