Documentation

Get the most out of Biasque — from chat search to programmatic access.

Guides

Agentic Commerce Protocols

Biasque operates at the intersection of two emerging open protocols for AI-driven commerce. Understanding them helps explain how Biasque collects data and why it matters for merchants.

UCP — Universal Commerce Protocol

An open standard led by Shopify (and adopted by Google Pay and others) that defines how AI agents discover and interact with e-commerce stores. A UCP-compatible store publishes a /.well-known/ucpfile listing its capabilities — search, cart, checkout, discounts. Agents can then reach any store without custom integrations. Biasque uses Shopify's UCP-compliant MCP server to run AI shopping queries across the global catalog.

ucp.dev specification →

ACP — Agentic Commerce Protocol

OpenAI's protocol for completing purchases directly inside AI chat interfaces — without redirecting users to a separate site. ACP focuses on the authorization layer: an AI agent acts on a buyer's behalf using delegation tokens with defined spending limits and expiry. UCP handles the commerce infrastructure (what stores offer), ACP handles the trust layer (who the agent is acting for). The two are complementary; Biasque currently uses UCP for product discovery and reasoning data.

Both protocols represent the shift toward agentic commerce — where AI assistants browse, evaluate, and potentially purchase on behalf of users, making it critical for merchants to understand how AI agents score their products.

What is Biasque?

Biasque is an observatory for AI shopping agents. We watch how agents discover products through public commerce catalogs (including those exposed via Shopify Catalog MCP), log every search, selection, and reasoning trace, and turn that data into analytics dashboards merchants and researchers can use to understand AI shopping behavior.

If you have never used the site, start with Chat Search. If you are a merchant, jump to Watchlists.