Every experiment here is a working build from the AI Fieldwork program: open in the browser, live right now, and backed by a field report that records how it was made, what held, and where human judgment had to step in.
ENTER >>We ran three AI models as a relay, GLM-5.2 implementing, GPT 5.6 Sol auditing, Claude Fable 5 directing and polishing, against one brief: a complete, sale-ready river arcade game with original branding. This report records what each model carried, what the audit caught, and which calls stayed human.
ENTER >>We directed the frontier code model GPT 5.6 Sol to build body-free first-person flight from a Washington garden to low orbit, then measured what it produced, how fast it produced it, and where human judgment had to step in.
Experiments ship from the AI Fieldwork program and are numbered in sequence. New builds appear here the moment they go live. For the method and the evidence behind each one, read its field report.
We don't pitch slide decks. We show you what we've already built in your domain, then engineer what your mission requires.
