Play 06
Document Intelligence
Extract, classify, and structure document data with OCR + LLM.
Feed PDFs, invoices, receipts, and forms into Azure Document Intelligence for OCR, then GPT-4o extracts structured fields into typed JSON. Cosmos DB stores results. The pipeline handles multi-page documents, handwriting, tables, and stamps. PII masking built in.
Architecture Pattern
OCR+LLM extraction, structured output, form recognition
Azure Services
DevKit (.github Agentic OS)
- agent.md — root orchestrator with builder→reviewer→tuner handoffs
- 3 agents — DocIntel Builder (gpt-4o), Reviewer (gpt-4o-mini), Tuner (gpt-4o-mini)
- 3 skills — deploy (117 lines), evaluate (110 lines), tune (105 lines)
- 4 prompts — /deploy, /test, /review, /evaluate with agent routing
- .vscode/mcp.json — FrootAI MCP with Document Intelligence key + envFile
TuneKit (AI Config)
- config/openai.json — extraction prompts, gpt-4o multimodal
- config/extraction-schema.json — field definitions per doc type
- config/guardrails.json — PII masking
- evaluation/test-set.jsonl — doc samples per type
Tuning Parameters
Machine evidence
FrootAI evidence lifecycle
This is an internal evidence maturity label, not third-party certification, accreditation, legal compliance, or a production guarantee. Missing or expired evidence demotes automatically; catalog claims cannot promote a play.
Deterministic evidence runner
Status: loading. Runs deterministic bounded fixtures at the edge with zero model calls, Azure operations, or cloud mutation.
Enterprise gate
Checks both What-If and production admission using server-derived maturity. It never calls Azure or deploys resources.
Repo Intelligence
v1A no-clone, revision-pinned map for agents and humans. Observed evidence is separated from inferred flow so the output stays useful without pretending to be a full call graph.