What you need to know
to make a decision.

The effort depends on your data, number of users, model, operating environment and integrations. A proposal should distinguish setup, infrastructure, model usage and support. We start by discussing your task to provide a reliable estimate.

You do not need your own AI hardware for hosting in a German data centre. For a private cloud or your own servers, we coordinate computing power, storage and responsibilities with your IT team.

Yes. Mikis can run with local language models entirely within your controlled infrastructure. Document preparation, embeddings, retrieval, logs and backups are also set up internally. Confidential content then does not need to be transmitted to external AI providers. If you want to use external models or interfaces, we explicitly define the permitted data and connections.

A custom integration can provide information from business applications, databases or document management systems. We assess interfaces, permissions and update processes for each system. Automatic standard integration with every software product is not promised.

Define typical questions and tasks alongside professional assessment criteria. Compare search effort, answer quality and necessary corrections with your current process. Use the results to decide on further adoption.

No. Even convincing answers can be incorrect or incomplete. Responsible employees must review results professionally before use.

No. The purpose, your organisation's role, risk classification and actual operation determine the requirements. The technical platform is one part of this assessment.

Yes. E-learning or a tailored workshop can be the first step. Learning objectives, the target audience and required documentation are agreed in advance.