Data & Integrations

How the Brand Match App connects to external systems, what it reads and writes, and how data flows through the pipeline.

---

ca-data Backend

The Brand Match App has no direct database connection. All data is fetched from ca-data (https://ca-data.vercel.app), an internal Next.js API layer that reads from Neon Postgres. The app uses a server-side caDataFetch() helper for all API routes and Server Components.

Authentication: All ca-data requests include a Bearer token from the CA_DATA_TOKEN environment variable.

Caching: All ca-data calls use cache: 'no-store' — data is always fresh.

ca-data Endpoints Used

Endpoint	What the App Uses It For
GET /matches	All match data, filterable by agent/brand/creator/status/score
GET /match-stats	Aggregate stats for home dashboard cards
PATCH /matches/:id	Update pipeline status and notes on a match
GET /brand-intel	Brand intelligence for a specific brand
GET /brand-intel/available	Unclaimed brands list with last-contact data
GET /brand-intel/list	Brand list for dropdowns/search
GET /profiles	Creator profiles (rate, CPM, views, niches, platform)
GET /cross-sell	Cross-sell recommendations
GET /contracts	Contract lifecycle data
GET /conflicts	Creator conflict detections
GET /brand-emails/:brand	Latest email thread + AI summary for a brand
GET /fathom/:brand	Fathom call history for a brand
GET /non-roster/:brand	Non-roster creators sponsored by a brand
POST /claim	Assign an unclaimed brand to an agent

---

Monday.com

Monday.com is the source of truth for deal data and deliverable history. The app reads from Monday indirectly through ca-data, which syncs Monday data to Neon on a schedule.

What comes from Monday:

• total_deals per brand — COUNT of rows in monday_deliverables
• avg_rate per brand — AVG of creator_payment in monday_deliverables
• last_activity_date — most recent deal date per brand (synced every 6h by sync-brand-activity.js)
• Open deal pipeline on the Creator Dashboard

Deal name normalization: Deal names have renewal/extension/upsell/copy suffixes stripped (e.g. "SoFi Deal (renewal copy)" → "SoFi Deal") before matching to creator names or computing stats.

---

Close CRM

Close has replaced Copper as the primary CRM. Close contact and lead data is synced daily to the brand_intel Neon table.

Sync script: sync-close-contacts.js (cron: daily 4am UTC)

What syncs per brand:

• last_contacted_date — most recent outbound email timestamp
• last_contact_agent — the CA agent who sent it
• last_contact_subject — email subject line
• last_contact_summary — AI summary via gpt-4o-mini

Brand intel size: 2,786 brands in brand_intel, including 1,608 imported from Close leads.

API: Close REST API, key at .secrets/close_api_key.txt. Values are in cents (e.g. 4750000 = $47,500).

---

Airtable (Creators DB)

Creator headshots, niche tags, and contact info come from the Airtable Creators Database. The app reads this data via ca-data's creator sync, which pulls from Airtable on startup and every 2 hours.

Headshot images are served directly from Airtable's CDN via the URL stored in the creator's record. The app shows a placeholder if the headshot fails to load.

---

Fathom (Call Recordings)

Past sales calls with brands are surfaced on brand detail pages. Data comes from Fathom via the shared team API key.

ca-data syncs Fathom calls on demand via GET /api/fathom-brand. Summaries are stored in the brand_intel table's fathom_calls column.

---

Sponsor Detection Pipeline

The conflict_database is populated by a separate sponsor detection pipeline that monitors YouTube channel transcripts for brand mentions. Results are stored in Neon and served via GET /api/conflicts. If a creator in a match is actively working with the brand being pitched (or a direct competitor), a conflict warning badge appears in the UI.

---

OpenAI

Two features use OpenAI:

Feature	Model	Trigger
Pitch email generation	gpt-4o	Agent clicks Pitch button — POST /api/generate-pitch
Brand email summary	gpt-4o-mini	sync-close-contacts.js — run nightly

The OPENAI_API_KEY env var must be set in the Vercel project for pitch generation to work.

---

Brand DB (36K+ brands)

A separate dataset of 36,762 known brands (grown from 33,380 by a nightly merge cron at 3am CT). This is used for the category-gaps and agent-blind-spots features — it contains brands that sponsor in our creators' niches but are not CA clients. Separate from the 2,786-brand brand_intel table.

---

Env Vars

Variable	What
CA_DATA_URL	https://ca-data.vercel.app
CA_DATA_TOKEN	Bearer token for ca-data API
OPENAI_API_KEY	For pitch generation