In a strategy workshop a managing director recently asked us: “I don't want AI that impresses me. I want AI I can trust.” The term currently shaping the debate can hardly be summed up better: Intentional AI.

Intent instead of activism

Intentional AI is the counter-movement to the reflexive AI adoption of recent years. Instead of introducing tools because everyone is doing it, intentional AI starts with the question: which problem are we solving, for whom, and how will we know it is solved? Three properties form the core. Explainable: the system's decisions can be traced and justified — to employees as much as to customers or auditors. Collaborative: the AI complements human strengths instead of invisibly replacing work; the handover points between human and machine are designed deliberately. Controllable: at any time there is the possibility to intervene, correct and switch off — technically and organisationally.

Why now?

The timing is no coincidence. After two years of agent euphoria, cases are piling up where automation works but nobody can say any more why the system does what it does. At the same time, regulation and works agreements demand exactly this traceability. Intentional AI is therefore less a new technology than a quality standard — one that, incidentally, also measurably increases acceptance among staff. People prefer working with systems whose logic they understand and whose limits are clearly stated.

The practical test

Our simple test for every AI initiative: can the business department explain in two sentences what the system does and where its limits are? If not, the workforce is not the problem — the design is.

If you want to measure your AI initiatives against this standard, we are happy to support you: from the statement of intent to the system that delivers on it.