# The AI-Leveraged Company

**Track:** Scaling & Stewardship — AI for Entrepreneurship — complete (29)
**Framework / surface:** venture strategy
**Level:** Advanced
**Prerequisites:** When Your Customer Is an Agent

**In one line:** Automate, augment, supervise, own — deciding what humans do when software performs much of the knowledge work.

## Theory, aesthetics & inspiration

The AI-leveraged company is organized around a question older org design never had to ask: for each unit of knowledge work, should software do it, should software draft it for a human to finish, should a human supervise software doing it, or must a human own it outright? Automate, augment, supervise, own — the ladder every role decomposes onto, and the decomposition moves yearly as capability compounds, so it is a standing planning instrument rather than a one-time reorg. What survives from classic doctrine is Matthew Skelton and Manuel Pais's "Team Topologies" (2019): stream-aligned teams owning customer outcomes, platform teams reducing everyone else's cognitive load — with eval and data infrastructure now a platform concern as fundamental as CI. What is new is where humans concentrate: judgment, taste, trust, and customer contact — the edges of the work — which is why the forward-deployed engineer matters more than another layer of management. The leaders among AI-native companies run revenue-per-employee figures that make classic SaaS benchmarks look industrial; the discipline behind that number is asking, before every hire, what an agent plus the existing team cannot do — and whether that answer survives next year's model.

**Founder question:** For each role you plan to hire: automate, augment, supervise, or own — which is it?