AI Agent Development for Business – with Clear Permissions and Approvals

A self-contained agent connected to two or three systems typically falls under the Professional package from €8,000; simple assistant functions with a few tools can start in the Starter package from €2,500. Several cooperating agents in business-critical processes are priced individually as Enterprise. Usage-based costs of the AI provider come on top; the exact quote follows after the free initial consultation.

A chatbot answers questions in conversation. An AI agent pursues a goal and uses tools to do so: it queries systems, creates documents or creates records. Many projects begin as an assistant and are gradually extended into an agent.

A prototype for a clearly defined task is often ready after a few weeks. Integration, tests with edge cases and a pilot phase are added before productive use. You receive the concrete schedule after the analysis.

In most cases, yes, provided there is an interface, database access or another reliable way in. If an API is missing, we develop the connection ourselves. Whether that is technically and contractually permissible in a particular case, we check with you beforehand.

The agent receives only the permissions it needs for its task, and consequential actions such as sending or deleting are confirmed by a human. Termination rules and cost limits prevent endless loops, and every action is logged. Errors can't be ruled out entirely this way, but they can be detected early and contained.

Only the data needed for the respective step. We choose providers and settings under which data is not used for training according to the contract terms, with processing in the EU if you wish, and we document all data flows. This does not replace legal advice, but it makes your data protection review easier.