AI governance

AI you stay in control of, inside your PLM

Four questions to ask before adding AI to your PLM

Your PLM holds your formulas, your labels and your proof of compliance. Before adding AI to it, four questions decide everything.

Here are our answers about Chef AI, the beCPG AI module.

Dark chocolate éclair · Suggestions
Suggestions tab of the product record: Apply all button and a label compliance check proposed by Chef AI, with a button to accept itA suggestion that a user accepts or rejects
Calculations

1. Which calculations use AI?

None of those with legal consequences.

Nutritional values and Nutri-Score, ingredient and allergen labeling, QUID, CLP classification, costs and optimization under constraints: the beCPG engines calculate them from your formulas. Same data, same result, every time.

AI is used where a mistake costs a correction, not a product recall: reading documents, searching, writing, translating, proposing.

Nutrients · calculated by formulation
Nutrients calculated by formulation: value per 100 g, per serving, labeled value with tolerances and calculation methodCalculated by the beCPG engines from the formula
Data

2. Where does our data go?

Optional module

The AI module is optional: it is activated at your request.

Request processing

Once activated, Chef AI requests are processed by Google Gemini, the model Chef AI is built on, with the access rights of the user asking the question. The AI module stores no business data.

Hosting

beCPG does not use your data to train AI models. The beCPG platform itself runs on our SaaS in the European Union, or on your own servers.

Dark chocolate éclair · Suggestions
Chef AI suggestions for the dark chocolate éclair: label compliance check, allergens to add, and the number of tokens used with the estimated costSaved only after confirmation
Validation

3. Who decides what gets saved?

You do. In beCPG, every AI proposal is a suggestion that a user accepts or rejects. Nothing is saved in the PLM without a person’s confirmation.

This matters more than it seems: an assistant that reads supplier documents and customer specifications handles content you do not control. Human validation before any change limits that risk.

An external assistant connected through the MCP server works with the access rights of the signed-in user, according to the rules set for your instance.

Traceability

4. Can we prove what happened?

An accepted suggestion is traced like any other change in beCPG: what changed, who validated it and when, on top of versions and your approval workflows.

Chef AI also shows, under each answer, the beCPG tool used, the number of tokens and the cost.

Chef AI · Dark chocolate éclair
Chef AI answers a question about the allergens of the dark chocolate éclair, showing the tool used and the raw materials in the formulaThe beCPG tool used, shown in the answer
Summary

In one sentence

AI helps your teams work faster. It does not decide what you declare, it only saves what a person validates, and the calculations you are accountable for are still performed by the beCPG engines.

Send us your security questionnaire. We answer it before the demo, not after.

Security and platform

Ask us the hard questions

Send your security questionnaire or book a demo with a PLM expert.