Triage. Then the unsaid.
Three thesis cutoffs run in seconds. Then we read what the deck didn't say. Conviction starts in absence, not presence.
Agentic creation (voice → world) fits the natural-language-development frontier of THV's thesis.
WRAPPER RISK. Generative 3D + voice is one OpenAI demo away from commoditization. Moat is the engine + distribution, not the AI call.
PT-based 3D engineering talent at 0.32x SF cost. Founder is technically credible — not a moat, but enables longer runway.
The most expensive sentences in venture are the ones never written. Below: gaps in this deal that should bend conviction down or trigger a press call.
- No retention curve. Creator tools live or die on D7/D30 — its absence is loud.
- No mention of distribution strategy beyond 'organic creator community'.
- Engine architecture details collapsed into 'proprietary' — would need to verify under NDA.
Wrapper risk surfaced. Verdict held until tech deep-dive under NDA + retention numbers. Press call scheduled.
Live triage on a fresh deal.
Drop in a deck blurb, founder email, or company URL. The same three cutoffs run on it in real time. No black boxes — every score traces to its inputs.