# Finding the AI Wedge

**Track:** Foundations of the AI-Native Venture — AI for Entrepreneurship — complete (29)
**Framework / surface:** venture strategy
**Level:** Beginner
**Prerequisites:** Jobs-to-be-Done for Probabilistic Software

**In one line:** The beachhead where today’s reliability already clears the bar — chosen on a frontier that widens every release.

## Theory, aesthetics & inspiration

Geoffrey Moore's beachhead doctrine from "Crossing the Chasm" (1991) — dominate one narrow market completely before expanding — gains a new selection criterion when the product is probabilistic: the wedge is where today's reliability already clears the bar. Apply the failure calculus market by market: with agents still failing roughly one in three attempts on structured benchmarks (Stanford AI Index, 2026), the entry point is work where a two-thirds hit rate with cheap review is already ten times better than the status quo — not work that needs five nines the curve has not delivered. The AI-era amendment to Moore is that the beachhead sits on a moving coastline: capability jumps widen the set of jobs the product can hold, so choose a wedge on a widening frontier — where each release extends your reach into adjacent work — rather than a niche the next release lets anyone serve. The wedge is also a data position: small enough that your corrections corpus clears the local quality bar fast, specific enough that the flywheel it starts is yours.

**Founder question:** Where is today’s model already reliable enough to be ten times better than the status quo?