Seventeen agents run in our own production every day. This page shows what they do, which systems they touch, and what a human still decides. Grouped by tier so you can find your own case.
Book a first callMost AI agent example lists are wish lists. This one is not. Every example carries a label: the agent either runs in our own business, or it runs at a customer, or it is a pattern we have not built yet.
We group the examples by tier because that is the honest measure. The industry does not determine the effort. What determines it is how many systems the agent works across and how much judgement the task requires.
One system plus email, a fixed flow. The agent prepares, a human approves.
Two to four systems. The agent applies your rules and exceptions and escalates special cases.
Five or more systems or operation inside your own network. Open tasks requiring judgement, coordinating other agents.
That is normal. Every agent is built for one specific flow, and the list above is our own daily work plus the first customer case. The question in the first call is always the same: which recurring task takes the most time in your company each month?
The best fits are tasks that recur often, follow clear rules, and tie up people without demanding their judgement: reviewing documents, moving data, checking completeness, chasing what is missing, producing recurring reports. Poor fits are tasks where the rule only emerges in conversation, or where a mistake becomes expensive immediately without anyone seeing it first.
Two questions. How many systems must the agent work across, and how many exceptions are there to the rule? One system plus email with a fixed flow is an Assistant. Two to four systems with exceptions is a Specialist. Five or more systems, operation inside your own network, or tasks needing real judgement are an Expert.
Because we apply what we sell to ourselves first. The seventeen agents in our business are our test bed: what runs reliably there for six months is what we offer customers. What fails there, you hear about from us before you pay for it.
€199, €499, or €899 per agent per month depending on tier, plus a one-off setup fee based on the number of connected systems. Model and token costs, hosting, monitoring, and model upgrades are included.
It can, but only where you define that it may. Every example on this page states explicitly what a human decides. Critical steps go for approval, every action is logged, and you can stop the agent at any time.
Thirty minutes is enough to find out. If no agent is worth it, we will say so.
Book a first call