00:00:50 · Dimitris Samouris
[00:00:50] S1: The truth is we kind of fell into it. So, very atypical founder story. I think I... Taking a step back, my career in consulting was very much a love hate relationship. I loved the work we did. I loved the clients we serve. I did a lot of private equity work, extremely sharp people, very demanding, but very straightforward to to deal with. And my personality type was always very ADHD. I loved the fact that I could, you know, tackle a new market for three four weeks, get really really deep into it, and then you were done and you'd move on to to something else. The bits that I hated were frankly the donkey work. I'm sure as a former consultant yourself, you remember late nights locked in a basement, cleaning call notes, aligning boxes on a PowerPoint slide till o'clock in the morning. Those are the bits that I felt that AI could take away and leave people more time to think about the stuff that actually moves the dial, which is, you know, the big picture answer. But I didn't leave consulting to do that. And this is where I think our our story differs a little bit from, from the typical kind of founder's journey of, hey. Like, I felt this pain point and I really needed to go leave to go, solve it. I... Actually, when I left consulting, spring of twenty two, I was done. I had done ten years. I left right before partner. I was really excited to retrain as a software engineer. I went to a coding boot camp. I taught myself how to code. My co founder, and I'm sure we'll get into this, I I knew him for way back and we talked about doing a company. When we got into our accelerator, we had zero intentions of building for a consulting or private equity. Actually, I would say I was really excited to go do something else. And we spent the first couple of months looking at a bunch of different ideas, and then GPT three dropped in Christmas of twenty two, and that very quickly became the only thing we could think or or talk about. And it was very obvious that it was gonna completely upend the way that advisory work was being done. And so despite, I guess, trying in many ways to escape the industry, life is funny and it pulls you back sometimes. We went full circle and have just been working on this problem for the last three and a half or so years.
00:03:14 · Sebastian Johnson, Dimitris Samouris
[00:03:14] S0: Yeah. And it's interesting when that that moment kind of erupted into the world. It's sort of like your moat or your your experience becomes so valuable because you've seen this manual process, you've seen this manual work happen for ten years, and then it's almost like gives you the ability to see that new world [00:03:30] S1: Yes. [00:03:30] S0: Where you can take this technology and apply it in a different way. [00:03:33] S1: Yes. [00:03:34] S0: When did you realize that actually this
00:03:35 · Dimitris Samouris
[00:03:35] S1: could be used in consulting or private equity? Equity? This was mid December, mid to late December. Nico, my my co founder and I, we'd spent the last three and a bit months. She spent a lot of time looking at climate tech. We looked a lot at EdTech, accounting tech. Very frustrated. We couldn't get any traction anywhere. Felt like we were banging our heads against the wall. And Nick says to me, why don't we just do nothing over Christmas? Like, just reset for a little bit. He was also frustrated because he hadn't written code in months. So we were on the phone doing user research, talking to prospective customers, and he was really itching to get his hands dirty again. And, you know, the API came out and he built this app that... It's a Chrome plug in. You'd go on the BBC or the Guardian's website. You'd press a button and it would summarize it, the article in, like, three or four bullet points. And he sends it to me one day. I'm like, this is good. Like, these summaries are reflective of the underlying content. It's a good synopsis. I I would use this. And then it sort of dawned on me. I spent a lot of time running... Writing summaries consulting specifically for for expert calls when we're doing market research. The missing piece here of course being the transcription bit. So we would like, you know, spend the hour on the call furiously taking Verbatim notes at a 100 words per minute. You pay a lot of money for these calls and the content is very very valuable But we'd use transcription models at Parthenon for a long time. There are loads of vendors out there that offered... They're just really bad. Really bad. Like like, you could not rely on the underlying transcript. The word error rate was through the roof. Maybe ten days go by between Nick building the first version of this OpenAI drops Whisper. As a side project, by the way, for those who who don't remember this, go back and read the original blog post that the OpenAI team wrote. This explicitly said that we are not going to commercialize this. This was their 20% moonshot time or whatever, and Whisper just blew every other transcription model on the market out of the water. It was not even close, head and shoulders above everyone else. So, like, okay. Like, let's see how good this thing is. Let's let's record a couple conversations and the transcripts were great.
00:06:08 · Dimitris Samouris, Sebastian Johnson
And so that began this journey of like, what can we do if we stitch together speech to text and and language models? What what workflows in that space can we go after? And I think seeing the content produced by the first Whisper model was when I... It really hit me like, okay. The way that you run a diligence is gonna completely change over the next couple of years. [00:06:32] S0: And what was the first, you know, three, six months like? You know, did you get to a point where you'd built a solid product that was useful? Or how long did that take until you [00:06:41] S1: kind of [00:06:41] S0: had the sort of the wind behind your sails?
00:06:44 · Dimitris Samouris
[00:06:44] S1: I wanted to start building right away. My co founder was much more product minded than... He'd spent five years at Airbnb and then a couple of fintech companies afterwards. It was completely right to pump the brakes. He was like, I know you think you know the problem. Let's do a bunch of user research before we do anything. So the first like month and a bit was hitting up my network and asking people how they run expert calls and very very quickly realized it's quite consistent across the industry. We did something like 80 user research calls in the space that first month. Obviously dogfooding the tech and I was gonna say, making sure that like we were happy with the quality that it produced. Those conversations and a piece of advice for other entrepreneurs that are going down this journey, they became our design partners. Like, the people that were most excited when we were talking to them about this were the people that signed up immediately. And, you know, there's this cliche our accelerator kept telling us like when you when you hit product market fit, will feel it. Yeah. There's no like objective metric here that you're going for. Like you will just know when when people want to use your product. And and that's what happened like at some point we were like, I have a case starting on Monday. Can we go? Like, can I use this now? And that's when we started building. We lined up three boutique consulting firms, the ones that would give us the time of day early on, you know, the big enterprises they they... It's very very hard to serve them if you don't have a track record.
00:08:16 · Sebastian Johnson, Dimitris Samouris
[00:08:16] S0: Yeah. [00:08:17] S1: And we spent... This is like February to May, so three months or so building out like v one of the product where it had a couple of different phases. First phase is like, can we throw some old content on there and see how it works? Second phase, can we run this in parallel of an existing engagement? Because, you know, these people get paid a lot of money on live deals. Like, there is no room for error on this stuff. It has to work. You can't ask a consulting firm to switch over their workflows to, like, untested software. And at the end of the three months, those three customers signed up. They were happy with the quality of the of the product. And that's when we started going to market like May May twenty twenty three. [00:09:04] S0: That's amazing. What you say about product market fit remind me of what another founder said that it's... Before product market fit, you sort of feels like you're pushing a boulder up a hill. Yeah. And then once you've got product market fill... Fit, all of a sudden you're kind of chasing that boulder down the hill [00:09:18] S1: Yes. [00:09:18] S0: And you're just sort of running with it. Yeah. Was that... I mean, it sounds like you clearly wanted to be a founder. Right? Before you knew exactly what you're building, the thing that you wanted to do was to build something cool or great or useful. Yeah. What was that moment like where you're like, hold on. We've actually... We've got something now. We've just kind of struck gold. People are telling us they want this.