Independent review

A second model checks your AI's work. It is harsher than the builder on purpose: a claim without evidence does not count, and a plan whose steps can't be individually verified isn't a plan.

Note: unlike the Brain Pad, this sends what you paste to our server for the review. Don't paste secrets or customer data.

Paste the requirements and the evidence your AI produced — its requirement→evidence map, test output, artifact listing. Another AI model reviews each requirement against the evidence and returns PASS or FAIL with findings. Questions about which model does the checking? Email me.