From Sberbank to 500 Global: launching a funded AI startup in 3 years
No Bullsh!t, December 24, 2025, 84 min, 1,410 views. The guest is Serzh Sarkisyan, co-founder and CEO of TalkLabs, which builds an AI coach for corporate training programmes and raised money from the 500 Global accelerator.
| Published | |
|---|---|
| Length | 84 min |
| Views on YouTube | 1,410 |
| Language of the episode | Russian |
| Where to watch and listen | YouTube, Spotify, Telegram |
Date, length and view count according to data from the episode page on YouTube, as of the page update date shown in the footer.
In short
Before TalkLabs he worked at large Russian corporations and then launched his own projects: an AI photo application and a digital mortgage for digital nomads that failed because of the market cycle and a bet on his own money. The conversation runs in two blocks. First, what corporate experience gives a founder: domain expertise, access to data, resources for testing hypotheses and hiring, and practices for scaling a team. Then, in detail and from personal experience, the mechanics of a first-tier accelerator: what share they take, why a transparent jurisdiction and cap table matter, how due diligence works, what mentors and portfolio companies actually contribute, and why the accelerator's brand works as a mark of quality for investors.
Key ideas
- Corporate experience gives a founder domain expertise, access to exclusive data and reports, and resources for hiring and testing hypotheses that a small startup has nowhere to get.
- The guest's main personal failure was building on his own money and entering the market in the wrong cycle: a digital mortgage for digital nomads coincided with a fivefold rise in the refinancing rate and stopped adding up.
- Learning from other people's experience barely works, except for individual mentoring on a personal request, and that observation is where the product itself came from.
- Teams without scaling experience are better off bringing in an advisor who has taken a company from five to fifty people, or using a simple trick: copy a working unit instead of building a complex hierarchy.
- First-tier accelerators take between three and ten percent at pre-seed, and their value is not the money but the brand, a ready investor funnel, co-founder search, access to portfolio companies and an office hours network.
- Legal transparency is critical when entering an accelerator: the jurisdiction, who owns the intellectual property and how clean the cap table is directly decide whether the money arrives in days or months.
The full transcript
The episode is recorded in Russian, and the full transcript of the conversation is on the Russian page of this episode. English and Spanish subtitles for the strongest episodes are rolling out on YouTube in waves.
Open the transcript