BioCreative

Account & Contact Lifecycle

How BioCreative organizes, enriches, classifies, and pushes accounts and contacts to clients — from raw import in the Hub through the live outreach message generator. 4 phases, 20 stages, each traced to its actual file/function/RPC on disk. Click any stage to jump to its full reference card below.

Free / deterministic (SQL, regex)

Paid (Claude, Brave Search, or Clay)

Phase output / terminal

Phase A — Hub

Account Classification

Raw import → 6-stage enrichment/classification pipeline → ~$0.046/account total

Phase B — Hub → Spoke

Sync to Client Databases

ICP-filtered weekly cron pushes classified data into each client's spoke database

Phase C — Spoke

Contact Classification

Minimal-data contacts (name + title only) get tagged, filtered, enriched, and persona-classified

Phase D — Spoke

Outreach Messaging

The live generate-outreach-message Edge Function (v9.0) — positioning, live signals, matrix + fallback strategy, cached prompt assembly, Claude generation, telemetry, and delivery

Full Reference

What each stage does, what it costs, and what comes next

Phase A — Account Classification (Hub)

Phase B — Hub → Spoke Sync

Phase C — Contact Classification (Spoke)

Phase D — Outreach Messaging

Source of truth: _shared/reference/ACCOUNT_CONTACT_CLASSIFICATION_SYSTEM.md (v2.3) · Built with the html_diagrams skill v1.1 · Source: _shared/brands/biocreative/diagrams/workflows/account_contact_lifecycle.html