The teamspace service desk has one built-in AI feature today: condense ticket history. On request, it summarises a long ticket history with the current status, the decisions taken and the questions still open. That helps when handing over to a colleague covering for you and before calling the customer back. The feature is included in every edition, runs on the AI connection your company enters itself (endpoint, model, key) and is switched on by your administrator.
From the outside, the MCP server adds to this. In Claude or ChatGPT you ask, for example: "Which Bauer KG tickets are still open, and who is working on them?" This requires that a suitable MCP connection has been configured and enabled for you, and that you are allowed to see those tickets in teamspace yourself. Nothing goes through MCP that you could not do in the service desk yourself.
Further AI features for the service desk, a ticket dispatcher and a ticket prioritiser, are in the works and not yet available. Until then, channels and escalation rules share out the work. teamspace does not run a language model of its own, and ticket content is not used to train AI models.