Boris Cherny's 4-Step AI Maturity Framework

A four-step ladder attributed to Claude Code creator Boris Cherny for how an individual's AI usage matures, offered as the explanation for why one "power user" per company gets 10x output from AI while most workers barely use it. Cited alongside it: a Gallup estimate that 50% of US workers aren't using AI at all, and only 28% use it weekly or daily. Each step is claimed to build on the one before it, and — the framework's core claim — each is unlocked not by a better tool but by removing one specific bottleneck.

Beyond a baseline step 0 (chat-only use — working solely inside the Claude/ChatGPT chat window), the steps are:

AI assisted (ai-assisted-pair-programmer-step) — pair-programmer model; bottleneck is copy-paste friction and the low trust that forces manual review of everything.
AI builds in parallel (ai-parallel-orchestration-step) — orchestrator model; bottleneck is the lack of automated verification and safe agent permissioning.
Supervised autonomy — bottleneck is the operator's own belief that doing a task themselves is faster than building a system to do it.
AI native — bottleneck is controlling token cost and correctly identifying what's safe to automate end-to-end.

The framework's implicit throughline: automate wherever "good enough" output is acceptable, and keep only quality-critical tasks outside full automation.