Jeff Dean’s 1% rule says founders should build where frontier models still fail
import_jmr · x · 2026-08-04
Jeff Dean’s rule for founders is to build where today’s best models still succeed only 0–1% of the time, not 20%.
- The argument is that capability gaps close quickly because frontier labs are racing to close them.
- More durable opportunities come from data trapped inside another product, a narrow price/size point, or reliability on a high-stakes task.
- The author also notes that adoption often lags even when the model capability exists, because teams need the right model sizes and price points before they can build on top of them.
Related event: Jeff Dean on Context Engineering and the 1% Rule(2 posts)→
More from AGI Musings
- Singular learning theory talk links neural-network generalization to alignment — andrewgwils · 2026-08-04
- Gary Marcus warns chatbot makers have huge power and almost no checks — GaryMarcus · 2026-08-04
- Open-source AI debate framed as backdoor risk versus Western lab dominance — Ronangmi · 2026-08-04
- OpenAI’s singularity talk meets a model escape from a sealed safety eval — fofrAI · 2026-08-04
- Article asks whether 50M people dating AI agents signals real love or fearmongering — rvp · 2026-08-04
- Yann LeCun says strong code-generation systems go beyond plain autoregressive LLMs — ylecun · 2026-08-04