Local AI
The model runs on your own hardware. Prompts and documents stay inside your own infrastructure. Makes sense for sensitive company data and tasks that a local model can solve reliably.
See local AI
Orientation
Before you build an AI agent
Not every process needs AI. Sometimes classic automation is faster, cheaper and more reliable.
The AI check helps you see in a few minutes whether AI makes sense for your use case, which data is involved and where human oversight still has to stay.
The check provides a first orientation. It does not replace legal advice or a review by a data protection officer or correspondingly qualified legal counsel.
What the check looks at
- What data is processed
- How strongly people stay involved in the process
- Whether AI prepares decisions or influences them on its own
- What requirements for oversight and documentation arise
- Whether AI is the right solution for the process at all
Takes about two minutes. Your answers stay in the browser and are not stored.
Details for those interested: assessment path and background
Method
Orientation on the requirements of the EU AI Act
The check follows important questions of the EU AI Act. It starts with the kind of use, then possible special requirements, then transparency, oversight and documentation. The result is not a formal classification under the EU AI Act and not a legal assessment. It only shows which topics may matter in a later review.
- What kind of AI use is this? The first look is at what the AI is actually meant to do. Does it only support staff, or does it influence decisions about people?
- Is there an especially sensitive use area? That includes, for example, certain uses in HR, education, critical areas or biometric systems. The concrete use case always decides.
- What role does the human play? An AI that drafts and waits for approval is assessed differently from a system that executes decisions automatically or scores people.
- What data is processed? Personal data, health data, finance data or confidential company information need special attention. Access rights and storage locations belong with that.
- What oversight remains necessary? Even without a high-risk use, requirements can remain for transparency, AI literacy, data protection, documentation or human oversight.
The concrete classification under the EU AI Act depends on the actual use, the role of the company and further circumstances. The online check therefore does not replace a formal review.
Architecture
Local or cloud, depending on data risk
The check first answers whether and how AI can be used in a sensible way.
Only then comes the technical question of where the model runs. For sensitive data, local AI can make sense. For more demanding tasks a controlled cloud setup can be the better choice.
Cloud AI
The model is wired in via a controlled API. Contract, data processing and region are chosen to fit the use. Makes sense for complex tasks, large document sets or processes that need more capable models.
See cloud AI
Delivery
A good AI solution does not start with the model
The most important question is not which model is used. What counts is how the whole process works.
That is why I plan AI solutions from the start with clear ownership and traceable flows. Depending on the use case that includes audit logs, approvals, roles and permissions, version history, and documentation of the models, agents and tool calls in use.
AI should not simply take decisions and then keep running unwatched. Critical processes need defined limits and a clear path for human oversight.
Unsure how AI fits your process?
You know automation could help, but you are not sure whether AI is needed for it?
Then I walk through the concrete process with you. I check whether classic automation, AI or a mix of both makes sense.
The end result should not be the newest tech, but a setup that runs reliably, fits the data risk and can be operated day to day.