What problem does it solve? Lab results arrive as PDFs from different providers (LabCorp, Quest, hospital labs) with inconsistent marker names and units, making it impossible to build a single longitudinal history per biomarker without manual transcription and unit conversion. ## Core Features & Use Cases - Dual-pass PDF extraction: Transcribe each lab report twice independently, then cross-check the passes with the fulcra-labs engine to catch transcription errors before anything is written. - Deterministic normalization: The engine resolves marker aliases, converts units to canonical forms, sanity-checks values against physiologic bounds, and computes idempotent source ids so re-runs are safe no-ops. - Verify-before-ingest workflow: Every row gets an ok/review/reject verdict in a dry run; nothing reaches Fulcra without passing validation or explicit operator confirmation. - Use Case: Hand the agent a LabCorp PDF and say "parse my labs" — LDL-C, HbA1c, and TSH results land on the same canonical tracks as previous Quest results, forming one queryable time series per marker. ## Quick Start Save your lab-report PDF locally and ask the agent to parse it into Fulcra tracks, then review the validation verdicts and confirm before ingestion.
