Training · Confident, capable, ready to evidence
Train your team to use AI with confidence.
Practical AI literacy sessions that also support your organisation's EU AI Act Article 4 responsibilities. Your team is already using Copilot, ChatGPT, Claude or Gemini. The question is whether they're using them well: sharper prompts, cleaner outputs, sensible habits around what stays inside the building. Sessions built for how your team actually works, with a attached: a practical, risk-aware programme tailored to the AI tools, roles and decisions in your organisation. No jargon. No panic. No slides about fines.
The Article 4 bit, briefly
Four things worth knowing. That's it.
Who it applies to
Any organisation whose staff use AI tools at work. Not just tech companies. SMEs are in scope too.
What 'AI literacy' means
Staff understand what the tool does, where it's risky, and when to double-check its output.
When it matters
Article 4 has quietly applied since 2 February 2025. National authorities begin supervising from 2 August 2026. Nothing dramatic, but it's worth having your training on paper.
What 'good' looks like
The duty is to take reasonable measures, not guarantee outcomes. A short policy, dated training records, a light audit trail. That's what you'll have to show.
Workshop tracks
Three tracks. Pick the one your team needs first.
AI literacy for every team
"Staff use ChatGPT, Copilot, Claude or Gemini already, but nobody's sure what's private, safe or allowed."
- Plain-English what/why/limits across Copilot, Copilot Studio, ChatGPT, Claude and Gemini.
- Redaction habits, one-page acceptable-use policy your team will actually follow.
- A defensible record of who was trained, on what and when, that helps you respond to Article 4 without an ISO project.
Getting real work done with AI
"Teams have licences but still copy-paste the same thing five ways. No shared patterns."
- A pattern library your team uses on Monday: email drafts, meeting notes, spreadsheet clean-up.
- When to reach for which tool, without picking sides.
- Small cohort, hands-on, working with your real (redacted) documents.
Building a first governed agent
"Everyone's heard about agents. Nobody's shipped one that a compliance person would sign off."
- One useful internal agent: Copilot Studio, ChatGPT Team or Claude Projects.
- Guardrails, a human-in-the-loop step, and a named owner.
- No code required. You leave with the thing running.
On Monday morning
What your team will actually be able to do.
"I paste customer emails into whichever AI is open at the time: desktop, phone, whatever."
One agreed pattern per task, redaction built in, one tool per pattern.
"We've bought three different AI tools and nobody knows which does what."
A single-page map: this tool for this job, this one for that one. Everyone reads from it.
"I think we're compliant? Nobody's actually written anything down."
A one-page policy, a dated training log, an incident channel. Contributes to evidencing reasonable measures under Article 4. Not a moment more.
Article 4 questions we get asked
What Irish SME owners and managers ask before booking training.
- What does EU AI Act Article 4 require of us?
- Article 4 asks any organisation whose staff use AI at work to make sure those people have sufficient AI literacy for their role. It is a duty to take reasonable measures, not to guarantee outcomes. In practice that means role-appropriate training, a short written policy, and a dated record of who was trained.
- Who on the team should attend?
- Anyone who uses Copilot, ChatGPT, Claude or Gemini for work, plus the manager who owns the policy. Sessions are split by depth rather than seniority: general staff get everyday use and red lines, while people building or approving AI output get the risk and review material.
- How long is the training and how is it delivered?
- A half-day core session, delivered on-site anywhere in Ireland or remotely, with optional shorter role-specific follow-ups. It uses your own tools and your own examples rather than generic slides, so people leave knowing what to do on Monday morning.
- What evidence are we left with afterwards?
- A dated attendance log naming who was trained and at what depth, the policy version they were trained on, and a short AI tool register. That set is what answers a client procurement questionnaire or a supervisory query, and it is usually asked for by large customers long before any regulator.
- Is this legal advice or certified compliance?
- No. Artellis supports AI literacy and governance implementation, and the training and templates are decision support your team reviews and approves. It is not legal advice and it does not guarantee regulatory compliance. Where you need a legal opinion, we say so plainly.
- When do we need this in place?
- Article 4 has applied since 2 February 2025, and national authorities begin supervising from 2 August 2026. Nothing dramatic happens on the date, but having training on paper before then is far cheaper than assembling it under a client or regulator deadline.
Delivered for an Irish energy retailer, Irish secondary schools, and professional teams across Dublin.
Get your team ready for Article 4.
Artellis supports AI-literacy and governance implementation. This is not legal advice and does not guarantee regulatory compliance.