SuggestAPI

SuggestAPI helps people find what matters faster. We’ve added MCP and WebMCP to our Search and Discovery Layer, so AI assistants can use our search engine in their workflows, or while helping visitors

Comment

Inspiration

AI agents are increasingly capable of helping people research products, compare options, and make purchasing decisions, but most ecommerce websites were designed exclusively for humans.

An agent visiting a storefront often has to scrape the website to get context. Interpret buttons and forms, scrape product information, or rely on a separate API that may not represent the experience available to the shopper.

We built SuggestAPI around a simple idea: merchants shouldn't have to rebuild their search and product discovery infrastructure for every new AI interface. We bring Agentic Commerce to any ecommerce platform without replatforming.

WebMCP gave us an opportunity to bring that idea directly into the browser.

Instead of making an agent guess how a storefront works, the website can explicitly expose the actions it supports as structured tools. Humans continue using the normal storefront while agents get a reliable interface to the same underlying product catalog.

What it does

SuggestAPI adds an agent-ready product discovery layer to existing e-commerce search infrastructure.

For our WebMCP implementation, a participating storefront exposes structured browser tools that an AI agent can discover and invoke while visiting the website.

Our demo provides four core commerce capabilities:

  • catalog_search — search a merchant's catalog using natural-language shopping intent
  • product_lookup — retrieve structured information about a specific product
  • compare_products — compare multiple products and their attributes
  • recommend_products — find products matching a shopper's requirements

This enables interactions such as:

"Find me waterproof trail running shoes under $150."

Instead of scraping the storefront or figuring out how to operate its search UI, the agent can call the catalog search tool directly.

The shopper can then ask:

"Compare the first two."

or:

"Which one would you recommend for wet trails?"

The agent can continue using the structured tools exposed by the page while the human remains in the same shopping experience.

The important part is that SuggestAPI doesn't require the merchant to replace their existing search engine, storefront, cart, or checkout.

SuggestAPI acts as an interoperability layer between the merchant's existing product discovery infrastructure and emerging agent protocols.

How we built it

We designed the WebMCP integration as a thin browser-native layer on top of SuggestAPI's existing Agent Gateway.

link

Architecture diagram is attached.

Built With

Share this project:

Updates