Anki is built and governed by John Conway. Not by a large team. Not by a lab. By one owner with final authority over a working system of AI engines that scout, draft, verify, and prepare — while he decides what is allowed to enter the world.
John is not a programmer in the traditional sense. That is not the weakness of the story; it is the point of it. Anki was built through disciplined collaboration with AI architects and builders, one verified step at a time. The foundation was not coding fluency. It was judgment, taste, persistence, and a written doctrine for how the work must be done.
Every claim needs evidence. Every action needs approval. Every piece of work earns its place — or it does not ship.
AI can write with confidence even when it is wrong. This system does not treat confidence as truth. AI can help find and shape the work, but every fact must come from verified real-world evidence. No source, no claim. No proof, no page. The rule came from experience: one invented email address bounced, and the system changed. Now guessing is not a flaw to manage — it is something the system is built to prevent.
This system is built to correct itself. When something fails, the fix becomes part of how the system works. A bad email created the evidence rule. A blocked owner action clarified the approval rule. The result is simple: facts need proof, and the owner remains in control.
Why Anki?
The name reaches back to Sumer, where An was heaven and Ki was earth. The first written records were made in clay to prove what happened. This system keeps the same promise in a modern form: every step recorded, every action accountable.
The full story is unfolding at Notes on a Build: the decisions, corrections, and lessons behind the system. For those ready to move from reading to using it, the founding pilot is open.