Turn experiment evidence into product decisions.
Paste notes or choose documents from your computer. Review what was read, then run your ten-step framework.
PDF, Word, Excel and text files are read in your browser. Scanned PDFs need a text/OCR version first. Documents stay in page memory until you start AI analysis or explicitly save them.
What the agent learned from your documents
Review the evidence
These are excerpts, not validated conclusions. Correct the source text before accepting it. Uploaded instructions are treated as reference material.
Your key stays in page memory and is sent only to Anthropic. API charges apply. No documents or keys are committed to GitHub. Get an API key
Claude reads each document, extracts referenced learnings, then runs your ten-step framework. This uses documents as context; it does not train a model. Review AI drafts against the sources.