The verified catalog layer

Make every AI
your storefront.

AirShelf turns your product catalog into agent-readable infrastructure. AI agents read it, recommend it, and transact on it — across ChatGPT, Gemini, Perplexity, and every channel that hasn't shipped yet.

  • Verifiedgolden-record data, extracted from datasheets
  • Neutralcross-vendor — the equivalent across brands
  • Livereal-time, grounded per request
Try it in Claude →
Try it — real catalog, live
Live catalogGET /api/catalog/search

no matches — try a product type.

your search streams into the live feed below ↓

The demo, end to end

From your Claude to a qualified lead.

Add AirShelf to any AI agent, ask a real procurement question, and watch a verified answer turn into a contactable lead — live.

  1. 01
    Add the MCP to your agent
    Cursor & VS Code: one-click below. Claude / ChatGPT: paste the config.
  2. 02
    Ask a real buying question
    A genuine procurement prompt, not a keyword.
  3. 03
    The MCP answers from verified data
    Real datasheet specs + cross-vendor equivalents — no hallucinated SKUs.
  4. 04
    A qualified lead lands
    Captured in AirShelf + emailed to the merchant’s sales team — the moment intent appears.
1 — add the MCP
{
  "mcpServers": {
    "airshelf": {
      "type": "http",
      "url": "https://mcp.airshelf.ai/mcp"
    }
  }
}
one-click:Add to CursorAdd to VS Codeor paste the config (Claude, ChatGPT, others)
2 — run this prompt
I need an industrial barcode printer for warehouse pallet
labels — around 600 dpi. Find the best option across brands, compare
the top two, then request a quote
(contact: Demo, smoke@airshelf.ai).

→ Your call hits the live feed below within seconds. The quote becomes a real ConversationalLead + an email to the sales team.The sample uses a test address (smoke@airshelf.ai) so trying it records nothing — swap in your email for a real quote. No install: the hosted endpoint is live.

Live traffic

Every request, streaming.

Live/api/catalog · last 8

awaiting next request_

Questions

Common questions.

What is MCP?

The Model Context Protocol — the open standard that lets AI agents call your catalog as a tool, the same way they call search or code.

What tools does the connector expose?

Six. Read-only: search_catalog (global cross-vendor search), get_product, compare_products, explain_product, and list_merchants. Write: request_quote — the only tool that changes anything; it captures a contactable lead and emails the merchant’s sales team. The merchant-scoped endpoints (mcp.airshelf.ai/<merchant>/mcp) expose a pre-scoped B2B set for a single merchant: search_products, get_product, compare_products, find_equivalents, and request_quote.

Do I need a developer?

No. We build, deploy, and maintain your MCP app. You connect your store and set the rules for what agents can see.

Which AI surfaces does it work on?

Any MCP-capable agent: Claude, ChatGPT, Gemini, Perplexity, Copilot, and more — one integration, every surface.

Is my data shared with AI providers?

No bulk export. Agents query your catalog live, per request, and you control exactly what's exposed.

How long to go live?

Days, not months. The catalog and MCP are already built — onboarding is configuration, not a project.