Developer API

Brand intelligence for
AI shopping agents

One MCP tool gives your agent brand context — voice, values, occasions, and buyer psychology. A real endpoint you connect to, not a claim that ChatGPT is already calling merchants.

See the response →View code examples

Four steps to brand-aware recommendations

01
🔗

STEP 01

Get the server URL

Each merchant can share an MCP endpoint and API key with agent developers they want to work with. Ready as a capability — not inbound traffic from ChatGPT or Copilot.

02
⚡

STEP 02

Connect your agent

Point your agent at the server URL. Streamable HTTP transport — stateless POST, no session management. It discovers the tool automatically.

03
🛠

STEP 03

Call the tool

Call get_brand_intelligence with no args to browse, or with a productRef for deep intelligence on a specific product.

04
✨

STEP 04

Cite in your recommendation

Use the brand intelligence to make recommendations your agent can actually explain — with real reasons, not generic summaries.

Response explorer

See exactly what your agent receives. Toggle between browse and product mode to explore the full response shape.

get_brand_intelligence({})
{
  "brand": {
    "voice": "Confident, understated, technically precise. Speaks to riders who value substance over hype.",
    "tone": "Direct and knowledgeable — like talking to a friend who actually rides",
    "values": [
      "Craftsmanship",
      "Performance-first design",
      "Sustainability",
      "Rider community"
    ],
    "messaging_pillars": [
      "Built for the mountain, not the marketing department",
      "Every board tells a story of where it was made to go",
      "Performance you can feel on the first turn"
    ],
    "brand_story": "Ridgeline was founded in 2018 by two backcountry riders who believed the industry had lost its way — prioritising graphics over geometry, hype over handling. Every board is shaped in British Columbia and tested in the conditions it was designed for."
  },
  "products": [
    {
      "productRef": "gid://shopify/Product/8847201837",
      "title": "Ridgeline Apex Pro 162",
      "short_description": "Directional freeride board built for aggressive backcountry riding in deep powder and steep terrain.",
      "who_its_for": "Advanced to expert riders who prioritise float and stability at speed"
    },
    {
      "productRef": "gid://shopify/Product/8847201952",
      "title": "Ridgeline Drift 155",
      "short_description": "Versatile all-mountain board with a playful flex pattern for riders progressing beyond intermediate.",
      "who_its_for": "Intermediate riders ready to push into more challenging terrain"
    },
    {
      "productRef": "gid://shopify/Product/8847202104",
      "title": "Ridgeline Summit Split 158",
      "short_description": "Splitboard engineered for long approaches and technical descents in remote backcountry.",
      "who_its_for": "Experienced tourers who need reliability and performance far from the resort"
    }
  ]
}

The productRef for each product is returned in browse mode — use it directly in your product mode call.

Connect the endpoint

One tool. Two modes. Works with any MCP-compatible agent framework.

// Browse mode — get brand context and full product catalogue
const browseResponse = await client.callTool("get_brand_intelligence", {});

// Extract a productRef from the browse response
const topProduct = browseResponse.products[0];
console.log(topProduct.productRef);
// → "gid://shopify/Product/8847201837"

// Product mode — get deep intelligence for that specific product
const productResponse = await client.callTool("get_brand_intelligence", {
  productRef: topProduct.productRef
});

// Use the intelligence in your recommendation
console.log(productResponse.product.tagline);
// → "The board that disappears beneath you — until you need it most"
Transport: Streamable HTTP (POST) · Stateless · Auth: Bearer token from merchant

Private beta

Building a shopping agent?

We're onboarding a small group of agent developers. Request access to work with merchant MCP endpoints — and get support from the team.

Request access

Limited spots · No credit card required