AI agents & Zen AI
12 AI agents, one job each. Your team decides what gets applied.
From sourcing missing specs to translation and catalog checks.
Agent changes wait for your approval by default.
How it works
Review first. Automate what earns trust.
From first run to auto-apply.
01Run
Pick a job and the products.
One product, a category, a family or the catalog in batches. Now, on a schedule, or when products change once you switch that on.
02Review
Approve or reject in bulk.
By default, suggestions wait in one queue, with source links for sourced values. Edit before approving; rejecting needs a comment.
03Trust
Let a trusted agent apply.
Switch on auto-apply per agent, above a confidence level you choose. It’s off by default, and locked fields and some changes still wait.
Agents
Meet the 12. See what each one does.
Create
Product sourcing Source links
Looks up missing specs on the web, with a source link for every value it suggests. Lookups need setup for your workspace.
Product enrichment
Drafts missing attributes and fuller descriptions for thin product pages.
Translation
Finds untranslated content and drafts copy for each market.
Asset intelligence
Flags missing, broken or low-resolution product images and drafts alt text.
Clean
Taxonomy
Proposes the right category for each product and flags conflicts.
Duplicate detection
Flags products that look like the same item entered twice.
Variant grouping
Spots products that should be variants of each other, orphaned variants and missing sizes or colors.
Compliance
Spots missing regulatory and policy information.
Sell
SEO optimization
Writes search-friendly titles, meta descriptions and bullets.
Digital Product Passport
Flags missing EU Digital Product Passport data on the products it checks, and fills only what your catalog already states.
Channel launch
Drafts missing channel-specific product fields and new marketplace feeds (column mapping, image sizes, file format) for review.
Pricing optimization Findings only
Checks prices against your pricing rules and the competitor prices you record, then flags issues with suggested prices for your team.
Beyond the 12
Make it yours.
Build your own agents
Start in Agent Studio from a template or blank. Test on a real product before anyone can publish it.
Chain agents into flows
Run steps in order or side by side. Start from a template, or describe a goal and review the draft.
Zen AI in the PIM
Ask about your catalog or have it write copy. It checks before overwriting, deleting or publishing.
Connect Claude or ChatGPT
Connect through MCP, the open standard AI assistants use to call tools. They act with your permissions.
Credit usage in view
Agents, Zen AI and image generation use credits. Admins see usage by model and feature.
A record of who decided
Each agent suggestion keeps who approved or rejected it. Restore products from version history.
Good to know
Straight answers.
Still curious? Email the founders
Do Zen AI and MCP edits go through the queue?
No. Zen AI and assistants connected through MCP make changes directly, with the signed-in user’s permissions. Zen AI checks with you before it overwrites content, deletes, publishes, imports, changes a status or tags products. Agent suggestions are what wait in the queue, by default. Product edits keep a version history you can restore from.
Do agents look at product images?
The built-in agents don’t. They work from file data such as dimensions, file type, links and image versions, not the picture itself. The Asset intelligence agent uses that data to flag missing, broken or low-resolution images, and drafts alt text where the product has an alt text or caption field.
Which AI models do agents use?
Models your organization has enabled. You choose a default model for agents, one for Zen AI and one for image generation. AI usage draws on your organization’s credits.
Which agents only flag issues?
Duplicate detection, Variant grouping and Pricing optimization report findings in Insights for your team to act on, with nothing to approve. The other agents suggest changes, which wait in the approval queue by default, and several also raise findings.
Are AI agents included?
Yes. The AI agents and Zen AI come with PIM Hub. AI usage runs on a credit allocation for your organization, sized in your quote, and admins can see usage by model and feature.
Works with
See the agents on your own products.
Once we connect, we load a sample of your products.
