What Icecat Learned From Letting Sales Colleagues Try AI

At Icecat, we’ve spent years building and refining validated product data. We know its value — for retailers, brands, and shoppers. But how does that knowledge translate when you put it directly into the hands of AI?

To find out, we launched the alpha version of the Icecat MCP server. MCP — the Model Context Protocol — is an open standard that lets AI assistants connect directly to external sources of truth. In our case: Icecat’s global catalogue of verified product data.

And instead of starting with developers, we gave it to our sales colleagues.

Learning by Doing

During our latest team event, sales team members — with no technical background — were given an API token, Claude Desktop, and some basic Node.js setup instructions.

The integration worked. At zero cost.

And suddenly, product comparisons were happening live, powered by Icecat’s verified data.

It wasn’t flawless. The setup felt a bit technical at first. But that was part of the lesson: AI agents can connect to trusted product data without expensive licenses, and even non-technical teams can get real value from it.

What Works Today

At this stage, the Icecat MCP server supports lookups by BrandName and ProductCode for Open Icecat Content.

That may sound simple, but it solves a big problem: hallucinations. Instead of guessing, AI assistants can now base answers on the same verified content Icecat provides to the market.

What We’re Learning for Tomorrow

Seeing colleagues explore MCP sparked ideas about what comes next. The backlog is already long, and includes:

Search by GTIN
Search by intent (“laptop for liberal arts student”)
Search compatible accessories
Images overview
Digital rights management insights
Content gap analysis
Average product health scores
Mapping brands and categories
Completeness analysis incl. assets
Per brand/category product overviews
Reference files (e.g., taxonomies)
XLS upload and download
User profiles
Search across manuals and rich media

The biggest lesson? Once people see validated data flowing into AI, the possibilities multiply fast.

A Confident Next Step in Our AI Journey

For Icecat, this isn’t about trial and error in the dark. We already know the value of product data and how AI can enhance tools like Icecat Studio, Brand Cloud, and Icecat PIM

With MCP, we’re expanding that value outward — enabling AI agents everywhere to learn from Icecat’s product knowledge.

This alpha is just the beginning. But what we’ve already learned: it works, it excites, and it opens the door to much more.