The EU AI Act makes deployers responsible for staff understanding the AI systems they use. Article 4 is the literacy clause: people touching these tools should know what they are and what can go wrong. When I tried to write that training from scratch, I had a list of roles and a vague sense of "we should probably cover bias." That's not a curriculum.
LiteracyKit is a tool I built for that specific job. You paste your staff roles and the AI systems they work with, and it drafts an Article 4 AI-literacy curriculum — modules mapped to the Act, plus the evidence gaps it can see in what you provided.
What the draft actually gives you
The output is module-shaped, not paragraph-shaped. Each module ties back to what Article 4 asks people to understand, and the tool keeps the focus on the Act rather than generic "AI is cool" material. For a deployer, that Act-focus is the point: the obligation is defined by the regulation, not by best-practice blogging.
The part I found most useful is the evidence gaps line. If you tell it "support staff use the chatbot" but nothing about how, the draft flags that it can't build a module on training that isn't described. It shows you the holes instead of filling them with plausible-sounding filler.
Where AI does belong in this picture
Drafting curriculum prose from a role list is open-ended enough that a model earns its place — there's no single correct wording. What it must not do is invent capabilities your team doesn't have. The tool stays in decision-support: it proposes modules, you decide what's real for your org.
What it deliberately does not do
- No legal advice. It drafts against Article 4; whether the curriculum satisfies an assessor is a human call.
- No claim your team is now compliant. Completing the modules is on you and your L&D process. The tool stops at the draft.
- No live tracking of who trained. It produces the plan, not the attendance sheet.
- No coverage of other articles. It's scoped to Article 4 literacy, not the rest of the Act.
LiteracyKit runs in the browser for $19/mo. If a module it proposed doesn't fit your roles, that mismatch is exactly the feedback that sharpens the next draft.
Top comments (0)