What we do/In depth
For IT leads and architectsHow it's built
How an AI agent is connected to your systems, where it runs, and how you stop or replace it.
A1 · The pieces
Your systems, the agent, and a person in between
The agent works through the same systems your staff use, with its own account and only the rights its task needs. Anything above its limits goes to a person.
Your systems
The agent
People
Log · every step the agent takes: what it read, what it did, which rule applied, who approved
A2–A6 · The details
Hosting, models, access and exit
Hosting and data
Data is stored in Switzerland. [provider, region]
Choice of model
The language model is the part that reads and writes text. It sits behind one interface and can be replaced. Which model a task gets depends on quality, cost and where the data may go.
Connecting to your systems
Each agent gets its own account in each system, with only the rights its task needs; read-only where reading is enough. Where a system offers no interface, the analysis says so before anything is built.
Monitoring and the off switch
A supervisor can pause or stop an agent at any time. Stopping hands the task back to people; nothing waits on the agent.
Leaving us
Configuration, rules, logs and the task list belong to you. The way back, to your own team or to another provider, is written down on the first day.
NextWho answers for what, and which rules we map to · Control and law →