Auto Store Call
AI phones stores, asks what's on the shelf, and turns the answer into structured signals.
An AI product suite for the beverage industry
Clinq Labs' discovery portfolio isn't eight disconnected tools — it's a single pipeline that mirrors a wine going to market: sense demand, manage price, win the placement, work the field. Each product feeds the next.
The architecture
Four functional layers, read like a wine being made. Signals flow down — what's selling and where it's about to run out triggers everything below it. Select any product to see what it consumes and what it produces.
AI phones stores, asks what's on the shelf, and turns the answer into structured signals.
Projects out-of-stock dates and reorder timing; owns the distributor-level Delta inventory model.
Replaces Sam's 253-sheet pricing spreadsheet. The one product that turns pricing into a queryable service — a price-lookup API and an RFP feed.
Auto-fills retail-chain RFP templates in place, grounded in real pricing — with a mandatory pricing-approval gate.
Turns an account into co-branded decks and sell sheets in seconds, with auto-sourced property photos.
Runs a brand team through writing an agency brief live in their group chat. A marketing-side island.
Plans a rep's day, pulls the right collateral, scans the shelf for gaps, and captures the visit by voice.
The system of record. Captures visits by voice, files photos, and answers questions over the whole book.
Nearly every dependency traces back to two master files and a common vocabulary. Today it's files; the Price Suite is what turns the spine into a service.
The wiring
Every cross-product feed below is documented in a product's PRD — these are the contracts that drive build sequencing, not aspirational arrows.
| Feed | Interface | What it carries |
|---|---|---|
| Auto Store Call → Depletion | store_signal_feed | Per account × SKU: in-stock, shelf price, placement, last delivery. Store → distributor, one way. |
| Auto Store Call → Enablement | /api/routing/queue | The ranked "show up here today" list, with reason codes. |
| Price Suite → RFP Builder | RFP feed | Per-state, per-deal-level prices. The single most-cited cross-product feed. |
| Sales Deck → RFP Builder | SKU + pricing | Sales Deck feeds RFP but never owns the form-fill itself. |
| Enablement ↔ CRM | account book | ACT reads KNOW's records and NL-query surface. One direction. |
| CRM ⟵ forked ⟵ Auto Call Svc | codebase | Build-time fork: voice, extractor, scheduler, Account model. ~1:1 architecture. |
The plan
The map isn't just descriptive — it dictates a build order. Five decisions fall straight out of the wiring.
The Price Suite and RFP Builder both need the 253-tab pricing parse — RFP Builder's highest-risk component. Build it in the Price Suite, expose the RFP feed, don't duplicate.
Two products wait on store_signal_feed and the routing queue. Those API surfaces gate both depletion enrichment and field routing.
Enablement (ACT) reads CRM's (KNOW) account book. Fork CRM first, or stub the KNOW interface so ACT can build against it.
Three products parse the same master files today. A shared ingestion service removes the most duplicated, highest-risk work in the suite.
Zero runtime coupling means no waiting. It can be staffed and shipped entirely independently.
Already standing
This isn't a greenfield bet from zero. One product is already in production, and it doubles as the fork-base for another — with shared infrastructure running underneath the whole suite.
FastAPI + Arq, Postgres · Redis · R2, Retell + Claude. ~836 tests. The engine for product 02 (Auto Store Call) and the fork-base for product 08 (CRM Assistant).
A common object store for recordings, transcripts and photos; Anthropic Claude is the shared model across all eight products.