# Co-Determination Negotiations for AI Agent Rollouts: When Finnish Law Requires You to Negotiate

> Finnish co-determination law can require negotiation when an AI agent monitors employees. Here is how growth companies run the process before rollout.

- Published: 2026-09-08
- Author: Master Mind
- Canonical: https://aimaster.fi/en/artikkelit/yhteistoimintaneuvottelut-ai-agentin-kayttoonotossa

Most growth companies roll out an AI agent like any other software update. The law disagrees the moment that agent starts tracking what an employee does, how fast they work, or what they say. At that point it stops being an IT project and becomes a change covered by Finland's co-determination law.

Many CEOs only discover this when staff ask who can see the logs the agent collects. Finland's Co-determination Act (Yhteistoimintalaki, 1333/2021) and the Act on the Protection of Privacy in Working Life (759/2004) place specific obligations on employers who introduce technical monitoring at work. An AI agent that reads emails, tracks working hours, or scores customer service quality meets that definition easily.

The same logic applies even when the agent looks like a small technical add-on. As an agent moves from pilot to production, its scope often expands quietly — a customer service agent starts reporting per-employee metrics to management. That is exactly the point where many companies cross the co-determination threshold without noticing.

## When does an AI agent become technical monitoring under Finnish co-determination law?

An AI agent becomes technical monitoring once it collects or processes information about what a specific employee does: working hours, messages, calls, or output. The Act on the Protection of Privacy in Working Life (759/2004) regulates exactly this — technical monitoring at work and the retrieval or opening of an employee's email messages. What matters is the purpose, not the agent's product name.

The line is fine in practice. An agent that answers customer questions does not, on its own, trigger the obligation. An agent that also measures how fast each named employee responded and reports it to a manager does. The same applies to an agent that scores a salesperson's call content or measures an individual's response times.

## What does the Co-determination Act (1333/2021) require from a growth company before rollout?

The purpose of the Co-determination Act is to guarantee employees timely information and a genuine chance to influence decisions that affect their work or position. The law applies to companies that meet the personnel threshold defined in the act itself — check your company's specific situation against the law text (1333/2021, Section 2). Once that threshold is met, introducing technical monitoring must be handled through co-determination before the decision, not after it.

In practice this means a growth company cannot announce an agent rollout as a finished package in an all-hands meeting. The negotiation has to happen while the scope and purpose of the monitoring are still open — not once the agent is already in production and staff notice its effects themselves.

## How does the Privacy in Working Life Act (759/2004) complement co-determination?

The Act on the Protection of Privacy in Working Life regulates technical monitoring at work and the retrieval or opening of an employee's email. It does not replace the Co-determination Act — it defines what monitoring may be agreed on at all. Co-determination defines how the decision is reached; the privacy act defines what the outcome may contain. A growth company should handle both laws in the same project, not split between IT and HR.

## How does a co-determination negotiation actually run for an AI agent rollout?

The negotiation is not a formality to survive before launch. It is a chance to identify, in advance, which feature of the agent will cause resistance — while the change is still cheap to make.

- Describe the agent's task and exactly what data it collects about an individual employee
- Identify whether the agent tracks performance, working hours, or communication — if so, handle it through co-determination
- Give staff advance information on the purpose, rollout method, and use of the monitoring
- Hold the negotiation before the rollout decision, not to rubber-stamp it afterwards
- Document the outcome of the negotiation and the agent's final monitoring features
- Update staff guidance whenever the agent's scope expands later

## How does a traditional software update differ from a monitoring AI agent at rollout?

| Stage | Traditional software update | AI agent that monitors work |
| --- | --- | --- |
| Co-determination duty | Usually not triggered | Triggered once the agent tracks a named employee |
| Timing | Notice is enough | Negotiation happens before the decision |
| Documentation | User guide | Negotiation minutes plus a privacy notice |
| Ownership | IT department | HR and leadership together |

Timing is not a problem if the negotiation is built into the project from day one. The Co-determination Act does not require months of process — it requires that staff genuinely get to weigh in before the agent's monitoring features are locked in. In a growth company this typically fits inside one or two weeks when it is prepared properly in advance.

Picture a typical situation at a growth company: a Finnish B2B telecom operator wants to introduce a customer service agent that analyzes call tone and response speed. When employee representatives join a [Master Plan](https://aimaster.fi/tuotteet/master-plan) sprint before rollout and the agent's metrics are defined together, the negotiation duty does not slow the project down — it removes the disputes that would otherwise surface later, because the boundaries of monitoring were agreed in advance.

The same sequencing applies to larger agent projects. When an agent is rolled out gradually with a clear [escalation path](https://aimaster.fi/en/artikkelit/ihminen-mukana-silmukassa-nain-kasvuyritys-suunnittelee-ai-agentin-eskalaatiopol), it is also easier to define the employee's role alongside the agent through co-determination — instead of explaining it after the fact.

## Does a small growth company need to negotiate before introducing an AI agent?

It depends on the company's headcount and on whether the agent tracks a specific employee. The Co-determination Act (1333/2021) defines its scope precisely in Section 2 — check your own situation against the law text before making a rollout decision. Even below the threshold, it is worth informing staff openly about how the agent works.

## Does the co-determination duty apply to every AI agent?

No. The duty applies to agents that collect or process information about what a specific employee does — working hours, messages, or output. An agent that automates a process without tracking individuals typically does not trigger the same duty. The line should be drawn case by case, not by the agent's product name.

## What happens if a company rolls out a monitoring agent without negotiating?

The risk is a labor-law one: staff can challenge the introduction of the monitoring, forcing the company to reverse or change the agent's behavior after the fact. That costs more time and trust than the negotiation would have taken upfront, and it slows down the next agent project too.

## Who is responsible for organizing the co-determination negotiation in an AI project?

Responsibility is typically shared between HR and leadership, not the IT department. The owner of the AI project should bring HR in during the strategy phase, so monitoring-related questions are identified before technical decisions are made — not after rollout.

The first step is not a lawyer's phone number but clarity on what the agent actually measures. Master Plan is an AI strategy sprint that maps where AI creates the most value for your company — measured in euros — while flagging exactly where a co-determination duty may apply. Book a [free Master Mind analysis](https://aimaster.fi/analyysi) and find out where your company stands.

## Frequently asked questions

### Does a small growth company need to negotiate before introducing an AI agent?

It depends on the company's headcount and on whether the agent tracks a specific employee. The Co-determination Act (1333/2021) defines its scope precisely in Section 2 — check your own situation against the law text before making a rollout decision. Even below the threshold, it is worth informing staff openly about how the agent works.

### Does the co-determination duty apply to every AI agent?

No. The duty applies to agents that collect or process information about what a specific employee does — working hours, messages, or output. An agent that automates a process without tracking individuals typically does not trigger the same duty.

### What happens if a company rolls out a monitoring agent without negotiating?

The risk is a labor-law one: staff can challenge the introduction of the monitoring, forcing the company to reverse or change the agent's behavior after the fact. That costs more time and trust than the negotiation would have taken upfront.

### Who is responsible for organizing the co-determination negotiation in an AI project?

Responsibility is typically shared between HR and leadership, not the IT department. The owner of the AI project should bring HR in during the strategy phase, so monitoring-related questions are identified before technical decisions are made.