Why I'm building AgentGEO in the open

Solo dev here. Every GEO tool I tried wanted to be my analyst — proprietary visibility scores, sentiment grades, weekly PDFs — when what I actually wanted was the raw data, inside Claude Code.

So I'm launching the layer underneath: one MCP call → the raw answers ChatGPT, Perplexity, Gemini, AI Overview, AI Mode & Copilot actually return, with citations, in one schema. Your agent does the thinking.

Launching here in ~2 weeks. Follow this thread if you want a ping when it's live — and tell me: which AI surface do you most wish you could see into?