AI you can understand and evaluate.

An AI initiative does not start with choosing a model. It starts with work that could be improved: structuring information, assessing text, exploring scenarios or supporting a recurring process. First we establish what the application must achieve and how the outcome can be checked.

I help define a useful application, investigate its limitations and set up human review. The outcome must fit the task, the available information and the user's responsibilities.

Workshops and training focus on AI literacy and practical use: what a language model does, how to formulate an instruction, where errors arise and how to assess results critically. Familiar exercises show when AI helps and when human intervention is needed.

For implementation, I can support prompt design, workflow development and translation of an existing work process into a concrete approach. Review points and the role of employees remain explicit.

Within insurance, I can apply subject expertise to model and domain training, scenario design, evaluation and output QA. This includes checking justification, completeness, consistency and correct use of technical terminology.

We agree the assessment criteria in advance. A fluent answer is not necessarily a correct one. We also examine assumptions, sources, missing information and the consequences of errors.

AI assignments outside insurance are also possible. My deepest sector expertise is in insurance; in other domains we determine what additional expertise is needed. We agree what will be delivered, who will review it and how the application will be used in daily work.