3 ways AI agents are reshaping PIM in B2B

Every PIM platform we implement shipped significant agentic capabilities this year. Here are the three use cases we believe matter most in B2B companies.

PIM still matters. But two things have changed.

For years, a PIM’s job was clear: be the single source of truth for product data and push it to every channel.

That job still matters. But two things have changed.

First, AI is now doing a growing share of the product data work itself.

Second, AI is increasingly the reader of product data. Assistants answer buyers’ questions, and agents compare products and vendors on their behalf.

Here are the three use cases we believe matter most.

1. Supplier onboarding: from copy-paste to review-and-approve

Ask any product data team where their time goes, and the answer is usually supplier data: PDFs, datasheets, spreadsheets in every possible format, copied manually into the PIM attribute by attribute.

That is exactly where AI agents are earning their keep first.

  • Struct PIM now uses AI to read supplier PDFs, datasheets and websites and map the content to the right attributes in your schema. Extracted values land in a review queue with confidence scoring before anything is saved, and Struct estimates 80–90% less manual data entry per new SKU. Read more about AI in Struct PIM
  • Akeneo’s Autumn Release 2026 adds new Supplier Data Manager capabilities, with AI agents working across supplier management and product catalogues under human approvals and full traceability. Explore Akeneo’s Autumn Release 2026
  • Inriver has previewed document-driven onboarding alongside project-based data remediation. Read about Inriver’s Summer 2026 release


Where it gets interesting for B2B:
extraction is only half the job. A manufacturer or distributor also needs that data classified correctly: ETIM classes, GS1 attributes, NOBB requirements. An agent that turns a supplier datasheet into validated, classified, channel-ready data is a step change in time-to-market.

Designing the data model and validation rules that make this reliable is where we spend most of our time with clients.

2. Enrichment and localisation: AI as a governed step in the workflow

The first generation of AI in PIM was a “generate description” button. Useful, but it left humans orchestrating every step. The 2026 generation puts AI agents inside the workflow, triggered by rules, coordinated with people and governed like any other process.

  • Inriver’s Spring 2026 release introduced the Enhance Agent and Expression Assistant, adding governed, agent-driven automation to the product information flow alongside existing enrichment and translation agents. A visual workflow builder lets business users define workflows that coordinate people, systems and Inriver’s AI agents. Read Inriver’s Spring 2026 announcement
  • Akeneo launched Agentic Ziggy in July, an orchestration layer that coordinates specialised AI agents across the product data lifecycle, from data modelling and schema mapping to enrichment. Explore Akeneo’s Summer Release 2026
  • Struct PIM lets you trigger AI automatically through business rules – for example, removing the background on every product image as soon as it’s uploaded. Read about AI-powered business rules in Struct
  • Bluestone PIM ships dedicated agents for on-brand enrichment (AI Enrich), translation and localisation (AI Linguist), and detecting and resolving data-quality issues (AI Analyst). Meet Bluestone’s AI Agent


The real question isn’t technical.
All four platforms can generate and translate content. The hard decisions are about governance:

  • Which categories can AI publish without review?
  • What confidence threshold triggers a human check?
  • How do you protect tone of voice across 20 languages?

Getting these rules right is what separates an AI pilot from an AI operation.

3. PIM as the source for agentic commerce

This is the biggest shift, and the one most businesses are least prepared for. When an AI agent researches, compares or buys a product for a customer, it reads your product data. If that data is incomplete, ambiguous or locked behind a website built for humans, the agent picks someone else.

The emerging answer is the Model Context Protocol (MCP), a standard way for AI tools to connect to systems like your PIM.

  • Inriver’s MCP server lets manufacturers and brands expose product data in a format AI agents can consume directly across the commerce ecosystem. Learn how Inriver approaches AI-ready PIM
  • Bluestone PIM offers MCP connectivity in which agents work under the same access controls, audit trail and workflow rules as the rest of the platform, with every action logged. Explore Bluestone PIM MCP
  • Struct PIM enables external AI tools such as ChatGPT, Claude and Microsoft Copilot to read and write product data in the PIM, with governance and audit trails in place. Read Struct’s take on Google’s UCP
  • Akeneo has partnered with Stripe on the Stripe Agentic Commerce Suite, covering product discoverability by AI agents and simplified checkout. Read about Akeneo’s agentic platform


A connector is not a strategy.
Having an MCP server doesn’t make your data agent-ready. Agents need attributes that describe use cases, compatibility, dimensions, certifications and sustainability data in a structured, comparable form.

Marketing copy that works on a product page often gives an agent nothing to reason with. Agent-readiness starts in the data model, which is why it belongs in the PIM project, not after it.

"Now, AI just sweeps through the data"

What this means for your PIM roadmap

Three practical takeaways:

  1. Start where the manual work is. Supplier onboarding and enrichment deliver measurable ROI quickly and build your organisation’s confidence in AI.
  2. Design governance before you scale. Human-in-the-loop rules, audit trails and confidence thresholds are what make AI safe to run at catalogue scale.
  3. Treat AI agents as a channel. Audit your data model for agent-readiness the same way you would for a new marketplace.

As a Struct PIM Preferred Partner, winner of the Inriver EMEA Partner Innovation Award 2025, and certified partner of Akeneo and Bluestone PIM, we’re helping clients put all of this into production.

Our Agentic Lab is where we test what’s next.

Want to know how ready your product data is for AI agents?

Book a
PIM assessment call

Want to know how ready your product data is for AI agents?

Book a call with our AI and PIM expert, Sivert.