scott wu

Cognition's internal metrics show Devin writing 95% of code and 7x code shipment increase in 6 months, while enterprise customers see 11-12x usage growth, signaling the category has moved from experimentation to production-scale productivity transformation.

Scott Wu reports Cognition's code output grew 7x in six months and customer usage of Devin grew 11-12x, arguing the category is moving from near-zero adoption a year ago to every software engineer globally within 2-3 years, driven by model improvements and product maturation.

Devin and similar agents now deliver 6-12x productivity gains on legacy modernization, with autonomous vulnerability remediation at 70% and a shift from human-scoped to event-driven proactive engineering where agents initiate work from alerts and scanning tools.

Engineers transition from writing code to managing fleets of AI agents, with the 'chief engineer' model requiring full-stack understanding as abstraction boundaries dissolve, enabling 10x more time on creative problem-solving versus implementation grunt work.

As AI makes software creation drastically cheaper, Jevons paradox drives exponentially more software production rather than less spending, creating a deflationary cycle where digital problems get solved abundantly while returns shift to physical-world execution, and AI empowers individuals to launch businesses without specialized labor across legal, financial, and engineering functions.

Enterprises avoid locking into single model labs because model leadership rotates rapidly; they need partners who drive organizational change (team structure, planning, specs, design) not just token throughput, creating moat for independent application-layer companies.

Enterprises want model-neutral partners to avoid lock-in to a single model provider (since leadership changes rapidly) and need help with organizational change — team structure, planning, spec design, user research — not just tool deployment.

Government agencies (Army, Navy, Treasury, NASA JPL) are deploying AI coding agents through FedRAMP-authorized platforms and prime contractors like Palantir and Anduril to modernize COBOL and custom legacy systems, bypassing broken procurement incentives by letting agencies own IP of agent-generated code.