cashcrown // How we build it
Knowledge is stuck in PDFs and scans — nobody reads them, and manual field re-typing is slow and error-prone.
file in memory, magic-bytes, zero disk write
reads layout and content (vision + text)
summarizes and extracts fields; Gemini 3 for very long docs
questions about the document with source references
Upload a PDF — see a summary, extracted fields and Q&A (file in memory, never stored).
An uploaded PDF/image becomes a summary, extracted fields and Q&A with citations.
Upload (PDF/obraz) → Qwen3-VL → Gemma 3 → Q&A z cytatami