Back to blog

Which PIM Platforms Support MCP? A 2026 Comparison

Which PIM Platforms Support MCP? A 2026 Comparison

At least eight PIM and product data vendors had launched or announced MCP support by November 2026. These include Bluestone PIM, Akeneo, Pimcore, Sales Layer, Stibo Systems, Informatica, inriver and Salsify.

What separates them is what an AI agent can actually do. Some MCP servers let agents search product records. Others support updating attributes, managing assets and carrying out catalogue tasks. This guide compares the available capabilities and gives you four practical tests to run before choosing a vendor.

What Does MCP Support Mean for a PIM?

The Model Context Protocol (MCP) is an open standard that lets AI applications connect to external systems through structured tools.

In a PIM, an MCP server allows compatible AI assistants to retrieve product information and, where supported, perform actions such as updating attributes or managing product relationships.

MCP usually works through existing APIs. An API makes a function available; an MCP tool makes that function accessible to an AI agent. The MCP server still needs to expose the relevant operation, and the connected user must have permission to perform it.

This makes API architecture an important consideration. Bluestone PIM, for example, provides more than 1,200 API endpoints and uses the same APIs for its own interface. Its MCP server exposes selected product management operations to compatible AI clients.

Which PIM Platforms Support MCP in 2026?

Bluestone PIM, Akeneo, Pimcore and Sales Layer document MCP capabilities that include reading and writing product data. Other vendors offer more limited access or have announced MCP capabilities without publishing a detailed tool reference.

Last reviewed: 2026. The comparison is based on publicly available vendor documentation and announcements.

PIM vendor MCP capabilities Publicly documented status
Bluestone PIM Read and write operations covering products, attributes, catalogues, assets and related tasks Documented product MCP server
Akeneo Read and write access to supported product information objects Available; capabilities depend on configuration and permissions
Pimcore Data-object and asset operations, including changes requiring human approval Documented in the Pimcore Agent framework
Sales Layer Separate read-only and read/write access profiles Documented MCP server
Stibo Systems Access to governed master data, with capabilities depending on the MCP service Consumer MCP Server generally available in 2026
Informatica MCP services for master data and other data management functions Announced in May 2026
Inriver Querying and retrieving product information Introduced in September 2025
Salsify Announced an MCP layer for access to approved product attributes Announced in May 2026; detailed write capabilities not publicly established

The difference becomes clear when you give an AI agent a task. Say 200 products are missing care instructions:

  • A read-only agent can identify the affected SKUs and show you what's missing.

  • With write access, the agent can draft the instructions using available product data and, once approved, save them directly in the PIM.

Bluestone PIM goes further: its MCP server lets AI agents update product data, manage assets and handle publishing tasks, so teams can get more catalogue work done faster, with fewer manual steps and full control over changes.

How Can You Test a PIM Vendor's MCP Capabilities?

The most useful MCP evaluation happens in a live demo, using a sample of your own product catalogue.

Run these four checks:

Test What to ask What to look for
Write access Which product objects can the agent create or update? A documented list of supported operations, including any restrictions
Coverage Can the agent handle assets, relations and bulk product updates? Working demonstrations of the tasks your team needs
Permissions How does the agent authenticate, and how are changes tracked? User-based access controls, appropriate approval steps and traceable changes
Availability Is the MCP server included in our licence and ready for production use? Clear documentation of release status, setup requirements and costs

Want to put these four tests into practice? Sign up for Bluestone PIM's free MCP Trial Server and connect your AI assistant to see what it can do with product data. Test the capabilities yourself and see how Bluestone PIM works beyond a feature list.

MCP Server

Try Bluestone PIM MCP Server

Connect your AI agent to Bluestone PIM and see how it can work with your product data.

Thank you for submitting the form.

What Can AI Agents Do in a PIM Through MCP?

With a read-and-write MCP server, AI agents can take on everyday catalogue tasks, from finding missing information to updating product records and preparing changes for review.

For example, a product manager preparing an autumn collection could ask:

"Find every jacket without care instructions, draft the missing text using supplier specifications and send it for review."

Instead of exporting product data to spreadsheets and copying changes back into the PIM, the agent finds the affected products, prepares the missing information and, with the right permissions, saves the approved updates directly in the system.

Bluestone PIM makes this possible through its MCP server, giving AI agents access to product data and management tools under the user's existing permissions. Teams can complete supported tasks directly in their catalogue, without building a separate integration for every request. Teams can manage everyday catalogue tasks using simple instructions, without having to work through each step manually in the PIM interface.

Other tasks an AI agent can help with include:

  • Which products in the Norway catalogue fail the completeness check, and why?
  • Link these 40 spare parts to the pumps they fit, using existing product relationships.
  • Match the new lifestyle images to SKUs using their filenames or metadata, then attach them to the correct products.
  • Prepare the updated catalogue for publication.

The result? Less time spent checking records and moving data between systems, and more products ready for sale without the usual manual work.

Should MCP Support Influence Your PIM Decision?

Yes, but MCP support alone isn't enough to choose a PIM.

Start with the product data workflows your business needs. Then check whether an AI agent can access those functions through MCP, which permissions apply and how changes are reviewed.

An open protocol gives businesses more choice over compatible AI clients. It doesn't guarantee that every client will support the same operations or that switching between providers will require no configuration work.

The safer approach is to assess MCP alongside the PIM's APIs, data model, workflow controls and integration options.

A vendor should be able to demonstrate how an agent finds a product, proposes a change, receives approval and verifies the result. That tells you far more about its usefulness than the presence of an MCP logo.

For more questions to ask during vendor selection, read The PIM Buyer's Guide to Agentic Commerce Readiness or talk to our team.

Contact us

Want to see what an AI agent can do in your catalogue?

Book a meeting with the Bluestone PIM team!

Thank you for submitting the form. We will reach out to you within 24 hours.

 

FAQ

  • No. An API exposes functions and data to other software. MCP provides a standard way for AI applications to discover and use tools connected to those functions. In a PIM, the MCP server commonly uses existing APIs to retrieve or update product information.

  • Yes, if the MCP server provides write tools and the connected user has the necessary permissions. Supported operations vary by vendor. Some changes may need human approval before they are applied.

  • MCP-compatible clients include Claude, ChatGPT and certain development tools. Compatibility depends on the client's supported features, subscription plan and the server's authentication requirements. Check each combination before deployment.

  • MCP can support controlled access, but security depends on the implementation. Look for restricted permissions, secure authentication, approval before significant changes and ways to verify what the agent has done. Test write operations on non-production data before using them in live catalogues.