Step 2 of 4 · ProjectsAutomations
ExtractThinker
Document intelligence library that uses LLMs to extract structured data.
What it does
Python library that turns documents (PDFs, images, text, spreadsheets) into validated Pydantic objects: it loads, classifies and splits them, then extracts fields through a large language model (LLM). Needs Python 3.9 to 3.13 and an API key for an LLM provider, or a local Ollama model. Apache-2.0; scanned pages need OCR or a vision model, and a valid schema does not guarantee accuracy.
Where it fits on the map
- Company tax department › Controversy & Tax RiskTax controversypartly covered
- Company tax department › Tax Technology, Data & TransformationTax technology & datacovered
- Tax authority › Compliance Risk & AuditAudit & investigationpartly covered
- Tax authority › Data, IT & AnalyticsData, document AI & AI modelscovered
In the processes
- M&A taxStep 2 · Due diligence
Similar projects
Of the same kind, for a shared jurisdiction or tax area. Not a ranking.
- CACAO (IDB)IDB web platform for tax administrations to receive, validate, store and analyse taxpayers' accounting data.
- CBCR-TPChecks a country-by-country report XML file against the OECD schema, Luxembourg rules and safe harbour tests.
- Comparables_IdentificationScript that ranks candidate comparables by the similarity of their business descriptions to a tested party.
- CPAverse Tax Prep AgentAI agent for tax preparation workflow automation, with agent skills and standard operating procedures.
- crs-lintOffline linter for OECD CRS XML (v2.0, 3.0): schema plus 41 status-message business rules, SARIF output.
- crs-xml-converterBrowser tool that builds OECD CRS returns (XML v2.0 and v3.0) from a spreadsheet and keeps a filing ledger.
Maintain this project?
Link to this page from your README, so readers find the facts, the notes and where it fits. The badge is a small SVG served from this site.
[](https://tax-oss.com/projects/extractthinker/)