Role-based AI literacy workshops
Use actual tasks such as drafting client correspondence, summarising records and checking AI-generated analysis. Compare good outputs, misleading outputs and when a human must intervene.
Enablement & compliance
AI literacy is about the decisions people make with real systems. We turn your team's uses of AI into practical exercises, review rules and an evidence trail. EU AI Act Article 4 calls for literacy-support measures; it does not prescribe a universal proficiency score.
Discuss team trainingThe working exerciseWhat this includes
Built around your tools, roles and decision points rather than a generic lecture or a compliance badge.
Use actual tasks such as drafting client correspondence, summarising records and checking AI-generated analysis. Compare good outputs, misleading outputs and when a human must intervene.
Document who uses which tools, the purpose, data involved, review step and escalation owner. Flag questions for legal or compliance specialists instead of promising certification.
Create sample failure cases, simple checking routines and an escalation path. Adapt the materials for staff who use AI daily and managers who approve its use.
What you walk away with
An agreed session outline, practical scenarios and a way to revisit gaps as your use of AI changes.
An initial inventory template covering purpose, data, decision owner and human-review points, not a legal declaration of compliance.
Concrete examples of when to verify, stop or ask for a second person rather than blindly accepting an AI answer.
The working exercise
A short scenario shows what the team must be able to recognise and who should act. The examples below are training prompts, not a statement that every control is legally mandated.
| In the workflow | Question to practise | Useful evidence |
|---|---|---|
| 01Before the prompt | Which records and personal data can be used? | Approved inputs and a data-handling rule |
| 02When the answer arrives | What would make this plausible answer wrong? | Source checks and an example counter-case |
| 03Before action | Who signs off or overrides the result? | Named reviewer and stop conditions |
| 04After a mistake | How is a concern reported and corrected? | Simple escalation and learning log |
Practical questions
No universal exam or individual proficiency threshold is specified in the amended Article 4. The appropriate measures depend on people's experience, role and how AI is used.
Training alone cannot settle every AI Act or GDPR duty. We can map use cases, teach human checks and flag questions that require a legal or specialist assessment.
Bring a few real use cases, the roles involved and examples of where outputs need checking. Redact confidential data and do not send credentials in the initial brief.
If you need the next layer
Choose the specialist page that matches the next decision.
We can use it to judge whether training, a readiness review or a narrower operating guideline is the useful next step.
Discuss team readiness