Context
Founded Youpi, an AI-native product studio exploring how generative AI and real-time LLMs could reshape creation, communication, and personal performance. Raised from top investors and built a 12-person highly-calibrated team to launch 10+ AI consumer products.
What I built
Realm AI
A consistent-character social creation app for AI-powered visual stories. The product explored how generative models could help people create scenes, stories, and imagined moments beyond what cameras capture.
Synco AI
A real-time AI copilot combining streaming transcription and LLM reasoning to assist users during interviews and live conversations. The product explored how AI could support users in high-pressure live moments by listening, summarizing, reasoning, and suggesting helpful responses.
What was technically hard
- Building multimodal generation systems with consistent character identity across scenes, which required novel approaches to model conditioning and visual coherence
- Designing real-time AI copilot infrastructure where streaming transcription and LLM reasoning had to operate within the latency constraints of a live conversation
- Navigating the cost-latency-quality tradeoff triangle across 10+ product experiments, each with different inference profiles
What I learned
- Where AI creates real user value and where it does not. Capability alone does not make a product. Timing, trust, and user readiness matter as much as model quality.
- How to own the full founder stack: product strategy, ML/LLM architecture, mobile/web/backend engineering, pricing, growth experiments, fundraising, hiring, and team management
- Practical lessons on AI product-market fit, distribution, and early-stage execution that no amount of big-company experience can teach
- How to build and lead a small, high-caliber team where every person matters
Why it matters now
The Youpi chapter is one founder journey that produced hard-earned operating judgment. Those lessons are exactly what I bring to founders at Embedding VC: clear-eyed understanding of AI product timing, GTM reality, hiring traps, and the gap between a working demo and a working business.