Proof is a tool for agentic engineering: you direct an AI agent to do the work, then check what it produced before you keep it. The work is generated; the solution is engineered, by you. Every decision you make is recorded, with where it came from, so the record itself is the evidence.

Proof has two sides. Ideation comes before the run: you frame the problem, explore a few approaches in plain language, and choose one. Review comes after: Proof reads what the agent changed and hands it back, clearing the parts that are safe and holding only what needs a look, each with its evidence, so you can approve it.

Proving is the move at the center. A concept starts open, a proposal you are still working out. You turn it into knowledge by doing the work: gather the evidence for and against it, run an AI challenge that argues the other side, check it against the project's success criteria, then sign it off. Promoting a concept records who proved it and on what basis, so months later "why is this settled?" still has a real answer.

The two sides connect. A chosen approach in Ideation carries its concepts into the work. A problem you find in Review can open a new Ideation project to frame the fix. The loop runs both ways.

Underneath, Proof does the analysis so you only see what needs you. The rigor stays under the hood. The surface stays plain: describe your problem, and Proof helps you frame and explore it. Methods are applied for you, not handed to you as homework.

Throughout, every decision is recorded with its provenance: the idea chosen, the approach picked, the concept proved, the change approved, the issue raised and resolved. That record is walkable, defensible, and exportable on demand.