At Mention First, I built an engine that measures how businesses appear in ChatGPT, Gemini, Perplexity, Claude, and Grok answers. Its integrity layer uses 20 deterministic checks to block reports that cannot be verified against raw answers.
I shipped and sold the product, including an operator console, and ran 26 fully costed live audits averaging $1.64 each. In independent research, I conducted 87 controlled Claude Code runs to examine what makes AI coding agents choose a developer tool.

