Before starting an AI project, make the main prerequisites visible. Use these questions to find missing arrangements. A number of checked boxes does not guarantee that a project is suitable, safe or profitable.
Data
- Which documents or data does the task need?
- Are they current, accessible and usable?
- Are you allowed to process them for this purpose?
People and organisation
- Who owns the outcome?
- Which staff members will test the pilot and give feedback?
- Is time available for training, review and maintenance?
Technology
- Can the required systems be connected reliably?
- Are permissions and a fallback for outages arranged?
- Who maintains the solution after initial delivery?
Process and results
- Is the current process documented, including exceptions?
- Is there one concrete problem to improve?
- Can you compare time, quality and costs before and after the pilot?
Decide the next step
For each answer, mark what is proven and what remains an assumption. Resolve blockers around data and ownership first. Then choose a small pilot with an end date and a decision point. This turns the checklist into a work plan rather than an apparently objective score.
