Agentic Document Analyser
Articles 11 + 18InfrastructureConverts unstructured compliance documents — risk assessments, model cards, contracts, audit logs — into structured JSON using Vision-Language Models. Acts as the evidence processing layer for the AiExponent compliance toolchain. Feeds Article 11 technical documentation and Article 18 log preservation workflows.
Quick Start
bashdocker compose upFeatures
- Vision-Language Model (Qwen2-VL) for unified layout analysis and OCR in a single pass
- Detects and classifies document elements: text blocks, headings, tables, figures, form fields, signatures
- Returns precise bounding boxes for every detected element
- Parallel page processing for multi-page PDFs
- Structured JSON output consumable by downstream compliance tools
- Docker Compose deployment — four microservices, one command
EU AI Act Context
Structures unstructured compliance documents into machine-readable JSON for Article 11 technical documentation packages and Article 18 log preservation workflows. Does not implement Article 9 risk management — use RiskForge for that.
Known Limitations
- Requires Docker Compose; no standalone pip package available.
- Depends on Fireworks AI API key — no offline/local inference by default.
- No persistent storage; results are not retained between container restarts.
- No authentication on the /analyze endpoint — not suitable for public deployment without a reverse proxy.
- Alpha quality: no production hardening, rate limiting, or database backend yet.
For the most current status, see GitHub issues.
Contributing
Contributions are welcome — Apache 2.0 licensed. See the contributing guide and open issues.
License
Licensed under the Apache License 2.0.
The Compound Moat
One tool is a start. The chain is the moat.
Each AiExponent tool produces structured evidence the next tool consumes. Browse the full toolchain — from Article 5 screening through Article 72 post-market monitoring.
See all tools →