Keynote Panel Discussion: Signal vs. Noise: Betting on AI’s Next Wave
summary
Five VCs debate concrete signals: forks over stars and founders who show grit by doing 100 customer interviews, not vanity metrics; one firm reports one in four inception-stage deals becomes a unicorn and cites Reflection’s $2B raise at an $8B valuation as a conviction bet. They test Embed grant apps with a model but find prediction loses spike events. Recommended metrics: pull request cycle time and deploys/day. Big opportunities land in tool-to-tool use and RL, while pretraining needs architectural breakthroughs like linear-attention or diffusion-for-code to stay viable.
AI-written summary · about
chapters (7)
- 0:00Welcome
- 4:00VC metrics for startups
- 13:33Operating experience in VC
- 18:30Hustler VCs in AI
- 24:48Frontier labs and pre-training
- 29:24Conviction vs trends
- 33:15Closing
related talks
56mKeynote Panel: AI from Investment perspective
33mKeynote Panel AI from Investment perspective
59mPanel: Artificial Intelligence: Who is Winning?Panel d4
22mThe Next Unicorns: 7 Top AI startups from the HF0 Residency
04mKeynote: Framing the Frontier of Machine Intelligence
33mRethinking How Startups Are Built and Scaled
31mThe Next Tech Unicorn
47mKeynote: AI is having its moment ... again
