Tool · prioritisation
AI automation: where to start?
Choose your sector, set the intensity of repetitive and document-heavy tasks for each department, and get a priority matrix (quick wins vs major projects) plus the top 5 automations to launch.
Tool
1 · Your sector
2 · Intensity by department
Set each slider according to the time spent on automatable tasks, per staff member:
- 1–2 · residual administrative tasks: 2 to 5 h/month
- 3–4 · manual data entry, proofreading, recurring reporting: 10 to 20 h/month
- 5 · fully document-based process, constant copy-pasting: up to 40 h/month
3 · Prioritisation
Top 5 to launch
Method: saving = time spent × automation rate (time spent comes from your rating, the rate depends on the task). Indicative estimate per staff member, to be multiplied by your headcount and refined during an audit; it is not a commitment. A staff member works in a single department, so the savings of different departments do not add up: the overall figure is their average. Cases handling sensitive data require a sovereign architecture.
Reading
How to read these results
This audit does not replace a process analysis: it gives you an order of priority. Its value is to move past the opinion-based discussion, where every director believes their own department is the most affected, and put a comparable, quantified baseline on the table.
The calculation
The slider converts time spent on automatable work, spread across the department's use cases. The saving is that time multiplied by an automation rate specific to each task, never 100%.
Quick wins first
The top-left quadrant combines high impact and low effort. Starting there is not a compromise: it produces the internal proof you need to fund what comes next.
What it does not tell you
Not the quality of your documents, the maturity of your teams, or the regulatory constraints on a given use case. Those three are verified during the pilot, not in a spreadsheet.
Next steps
After the audit
Automation potential is not a project. Three steps separate this matrix from a decision: cost the return on investment with your own figures, check the regulatory risk level of the chosen use case, then test it on your real documents.
Our other tools cover the first two: the ROI simulator turns these hours into euros, and the AI Act self-assessment places your use case within the regulation's risk tiers. If the tasks you identified are mostly document-based, the article on sovereign RAG explains what actually determines answer quality.
Scope
Potential capacity, excluded from budget
Displayed hours are gross task-level potential, before adoption and review. Do not add overlapping tasks or treat time as eliminated payroll. Measure errors, citations, access control and review time in a pilot, then enter the measurements in the shared calculator.