Enterprise Document Intelligence Platform

Pipeline

Custom RAG orchestration — not a prompt wrapper.

Evaluation snapshot

Metrics written after real eval runs — see reports/EVALUATION_REPORT.md

Source: reports/metrics/eval_metrics.json (sample corpus eval).

Integrity

• Answers include citation structures (chunk IDs)
• Sample/public corpus only — not a production plant archive
• Extractive grounded mode by default; optional LLM via API key
• Same portfolio DNA as Predictive Maintenance platform

Run

set PYTHONPATH=src
python scripts/build_index_and_eval.py
uvicorn docintel.api.main:app --app-dir src --port 8000
# http://localhost:8000/docs