Are your agents actually making decisions, or just automating steps humans already defined?

A “submit order” button executes exactly the user’s decision. An agent told “optimise my travel expenses” makes the decisions that matter: which vendor, which flight, which payment. The user provided the goal; the agent chose the actions.

That distinction is where the value lives — and where the risk originates. Traditional software forwards intent. Agents create it. If your agents are running predefined workflows with no room to reason, plan, or adapt, you’re getting automation with extra steps. The upside of agency comes from letting agents make the calls that matter within boundaries you define.

The question to ask your team: for each agent in production, can it actually choose a different path based on what it finds? Or is it following a script?

Go deeper: Why Traditional IAM Breaks Down explores what changes when agents create intent rather than forwarding it.