Multi-agent
Upload documents (PDF, Word, Excel, Markdown, text, CSV, JSON, images) or paste HTTPS URLs. Click Run: agents extract text, then work in sequence — Source Analyst (per source), Corpus Synthesis (merged training document), Reference Index (structured JSON). Downloads appear below. Original files are removed from the server after extraction to save disk space; only the generated outputs are kept under .data/corpus-builder/outputs/.
Up to 24 sources total (files + URLs). Large PDFs and sheets are truncated per source for model limits.