Frequently asked questions.

An AI assistant is software that receives a task, works out the steps needed, uses tools (read an email, search a folder, run a calculation, write a document) and hands you a result. It works autonomously, but within the frame you have set.

No, not by default. Anything that cannot be undone, such as sending an email or writing into a tool, waits for a person’s decision. An administrator can choose to allow advance approvals on some tasks, but that is an explicit choice.

Each customer has a dedicated space, isolated from the others and hosted in Europe. The credentials for your tools are encrypted and never leave the platform.

You choose the AI provider, assistant by assistant. The AI accounts are in your name, and you see what each task costs.

Every run is recorded step by step: what the assistant read, which tools it called, what it asked to be approved, and the exact version of the assistant that was used. You can reread any past run.

No. You describe the task in your own words, we set up the assistant with you, and there is nothing to install. You only approve what matters: outgoing messages and entries in your tools.

That is what approval is for: nothing goes out until you have reviewed it. Every step is recorded, so you can trace where a mistake came from and correct the instructions.

It depends on the task and its volume. We discuss it once we understand your needs, and the running cost of each task stays visible in your dashboard.

A general-purpose assistant is usually not set up for your tools, does not keep a record of every step and does not require your approval before sending. Here, your space is dedicated and your data stays in Europe.

Write to us describing a task that comes up often at your company. We look together at whether an assistant can take it on and how to set it up.