dark factory
In plain wordsA way of building software in which AI agents write, test and ship all the code and no human writes or reads it, like a factory run with the lights off.
A software-development setup in which coding agentscoding agent, n.: An AI agent that writes and fixes software by itself: it reads a codebase, edits files, runs commands and tests, and keeps working until a feature is built or a bug is fixed. write, test and ship code with no human writing or reviewing it; people supply specifications and judge outcomes. The best-known example, StrongDM's 'Software Factory', barred its engineers from both writing and reviewing code. In place of review it relied on end-to-end usage 'scenarios' stored outside the codebase, like a holdout settest set, n.: A set of examples kept completely apart from training and used at the end to measure how well a model works on data it has never seen, like a sealed final exam., so that agents cannot tailor the code to them; on a probabilistic pass mark, 'satisfaction', the estimated share of observed runs that would leave a user content; and on agent-built working replicas of outside services such as Slack, Jira and Okta (a 'digital twin universe') for testing at scale. The team treated token spending as a gauge of ambition, suggesting that under US$1,000 a day per engineer meant the factory had room to improve.
In a dark factory, the humans write scenarios and specs; they never open a diff.
Origin
Borrowed from manufacturing, where a 'dark' or lights-out factory runs without workers on site. Dan Shapiro applied it to software as the highest of five levels of AI-assisted development in a January 2026 post; Simon Willison's February 2026 account of StrongDM's team spread it.
Usage note
Skeptics question whether unreviewed code can be trusted in security-critical systems and point to the token costs; proponents argue that rigorous external testing can substitute for line-by-line review.
Related
- coding agentcoding agent, n.: An AI agent that writes and fixes software by itself: it reads a codebase, edits files, runs commands and tests, and keeps working until a feature is built or a bug is fixed.
- agentic codingagentic coding, n.: Writing software by directing AI agents that plan, edit and test code themselves, with the human acting more as a reviewer and manager than as the person typing every line.
- vibe codingvibe coding, n.: Programming by conversation: you ask an AI for what you want, try out what it produces and ask again until it works, trusting the code without really reading it.
- digital twindigital twin, n.: A detailed virtual copy of a real machine, factory, robot or city, kept in sync with live sensor data and used to monitor it, test changes safely and predict problems.
- evaluationevaluation, n.: Testing an AI system to find out what it can and cannot do, how well, and how safely — from exam-style quizzes to hands-on trials and expert reviews.
- harnessharness, n.: The software around an AI model that gives it instructions and tools, carries out its actions and decides when it stops. The model is the engine; the harness is the rest of the car.
- AI factoryAI factory, n.: A data center built to turn electricity and data into AI models and AI answers at industrial scale; in Europe, also a public supercomputing hub that helps companies and researchers build AI.
Sources
- Willison, S. (2026). How StrongDM's AI team build serious software without even looking at the code. Simon Willison's Weblog, 7 February 2026. simonwillison.net/2026/Feb/7/software-factory
- Willison, S. (2026). The five levels (on Dan Shapiro's levels of AI-assisted development). Simon Willison's Weblog, 28 January 2026. simonwillison.net/2026/Jan/28/the-five-levels
Mentioned in
automationautomation, n.: Using machines or software to do work that people would otherwise do, from factory robots to programs that sort email or write reports., AI factoryAI factory, n.: A data center built to turn electricity and data into AI models and AI answers at industrial scale; in Europe, also a public supercomputing hub that helps companies and researchers build AI., Deep Blue (malaise)Deep Blue (malaise), n.: A feeling of gloom and dread among programmers as AI tools become able to do much of the work that once gave their jobs meaning., Ralph Wiggum loopRalph Wiggum loop, n.: A blunt way of using AI coding assistants: feed the same instructions to the AI again and again in an endless loop until the software gets built., spec-driven developmentspec-driven development, n.: Building software with AI coding agents by first writing a detailed description of what it should do, which the AI then follows to plan and write the code., vibe engineeringvibe engineering, n.: Building serious software with AI coding assistants while still taking full responsibility for its quality — the careful, professional opposite of vibe coding., vibe portingvibe porting, n.: Using AI coding agents to translate a whole software project into another programming language, usually checking the result against the original project's tests.