Consulting on AI at work
Rolling out an AI tool across a European workforce is not an IT project. In most of your countries it is a consultation event, and in some of them it needs agreement before anyone logs in.
The consultation wave nobody staffed for
Every meaningful AI deployment touches how people are managed, monitored or evaluated, and that is precisely the territory where European co-determination bites hardest. In Germany, §87 BetrVG gives the Betriebsrat co-determination over technical systems capable of monitoring performance or behaviour, which captures a great deal of what modern software does by default. France, the Netherlands, Italy and Spain each have their own route to the same conversation.
Layer the EU AI Act on top, with its obligations on deployers as well as providers, and the volume of consultation rises sharply at exactly the moment the business wants to move fastest.
The difficulty is rarely that the company is unwilling to consult. It is that a single AI rollout can trigger fifteen separate conversations with fifteen different bodies, each needing its own information pack, its own timetable and its own record, and the employee relations team is the same size it was last year.
What Graylark does
Model the rollout once, consult per country. Change Proposals treats an AI deployment like any other multi-country change: one proposal, per-country obligations, per-country owners and deadlines.
Know which bodies you are dealing with before you start. ER Landscape holds the representative bodies, entities and agreements across the group, so you can see immediately which sites have a works council and which existing agreements already cover technical systems or monitoring.
Keep the reasoning with the decision. Country assessments are threads, not fields, so the argument you made to a works council in March is still there when the same question arrives in October from another country.
Evidence that survives scrutiny. Every information pack, response and opinion is captured as the process runs, which is what an audit, a regulator or a dispute will ask for.
AI you can defend in the room. There is a credibility problem in using an opaque AI tool to manage consultation about AI. GrAI runs entirely inside your tenant. No external AI API calls in the customer data path, no training on your data, a mandatory output-safety layer on every call, a per-tenant audit log of every AI request, and a master off-switch. None of Graylark’s AI features are classified as high-risk under the EU AI Act risk framework, and no automated decision is made that produces a legal or similarly significant effect on an individual.
Who this is for
Employee relations and labour relations leaders, employment counsel and HR technology owners at multinationals deploying AI, analytics or workforce monitoring across European operations.
See how it runs
Book a demo and bring a real rollout. We will map the consultation obligations with you.
Related: Works council management, Cross-border restructuring consultation, Security and procurement.