Leverage
An AI innovation league for software organizations, where teams compete on verified outcomes, reuse, and cost discipline rather than token volume. Early build, directed rather than hand-written.
Leverage explores the operating loop most AI-analytics products skip: leaders publish a bounded, quality-gated mission; vendor usage, repository, CI, deploy, and incident events become normalized evidence; deterministic scoring rewards outcomes and external reuse while neutralizing spend-only behavior; and winning methods become versioned playbooks other teams can adopt. The current repository is an early full-stack build with eight surfaces, a Convex data model, and an optional Rust telemetry gateway. Its commits so far were produced by a coding agent under my direction, which is why it sits on the record as an early build and not as shipped work.
Record