Finance Tracker
AI extraction and OCR for PDF statements, with a review step before saving transactions.
PythonFastAPIOCRDocker
A self-hosted Python API that extracts statement text, falls back to OCR for scanned PDFs and sends the extracted content to a selected AI provider. Separate upload, review/confirmation and transaction-query endpoints keep extraction and persistence distinct.
What’s included
- Text-based PDF extraction and OCR fallback.
- OpenAI and Anthropic provider paths.
- Review and confirmation before transaction storage.
- Transaction queries and cached submissions.