badge-qualifier

# Badge Qualifier Transform raw booth conversation notes into a structured lead record — including tier, authority, fit, and next step — without inflating signals that aren't there. ## Workflow ### Step 1: Normalize Raw Input Accept any of these input formats: - Typed booth notes ("Spoke with Sarah at Acme, she asked about pricing for 5 lines") - Badge or business card OCR text (name, title, company, contact details) - Voice transcript or dictated summary - A mix of all three If the user pastes badge text only, treat it as **contact-only** — do not infer conversation depth that wasn't described. Extract and confirm these fields before proceeding: - **Contact name** (badge or notes; unknown if absent) - **Job title** (badge; unknown if absent) - **Company** (badge; unknown if absent) - **How contact was made** (scanned badge / brief chat / product demo / pricing discussion) If critical fields are missing and the user is in a live session, ask a single clarifying question. If processing in bulk, mark as `unknown` and continue. ### Step 2: Extract Structured Lead Facts From the normalized input, pull explicit facts — not inferences: | Field | Source | Rule | |-------|--------|------| | Name / Title / Company | Badge or notes | Transcribe exactly; mark as `unknown` if absent | | Email / Phone | Badge | Transcribe only if present; never fabricate | | **Need** | Conversation notes | Only quote if explicitly stated; otherwise `unknown` | | **Urgency** | Notes ("needs by Q3...