AI Literacy Training for Employees: A 30-Day Plan

AI literacy training teaches employees to use AI well in their own jobs: when to hand work to AI, how to instruct it, how to check what comes back and what must never go in. The training that works is short on theory and long on practice, built on tasks people already repeat every week and measured before and after.

I run these programs for teams in English and Spanish. This is the 30-day structure I use, plus what the EU AI Act asks of employers and the free resources worth using.

What AI literacy actually means at work#

AI literacy isn't knowing how a neural network works. For most employees it comes down to four questions they can answer confidently:

1. Is this a task AI should help with?
2. What does it need from me to do it well?
3. How do I know the output is right?
4. What data, decisions or messages should stay with a person?

Anthropic's free AI Fluency: Framework & Foundations course, developed with Prof. Joseph Feller of University College Cork and Prof. Rick Dakan of Ringling College, gives those skills names. It calls them the 4D framework:

I use the same four words with teams because they're easy to remember and hard to argue with.

Is AI literacy training required?#

In the European Union, yes, in a proportionate way. Article 4 of the EU AI Act requires providers and deployers of AI systems to take measures ensuring a sufficient level of AI literacy among their staff and others using AI systems on their behalf. It accounts for people's technical knowledge, experience, education and the context the systems are used in.

The obligation has applied since February 2, 2025. As Crowell & Moring explains, it doesn't require guaranteeing a specific level for each individual, and enforcement by national authorities started in August 2026.

Practical takeaway: keep a simple record of what training you ran, for whom and on which tools. It's good management anyway.

The 30-day AI literacy training plan#

Week 1: Baseline and guardrails#

• List the tasks. Each person names three tasks they repeat weekly: reports, emails, research, spreadsheets, meeting notes.
• Time them. Record how long each takes today. This is your before number.
• Set the guardrails. One page: approved tools, data that never goes into AI, and when a human must review.
• Pick champions. One or two people per team who get extra support and answer questions.

Week 2: Core skills, in about four hours#

• Run a live session on the 4D framework using examples from your own company.
• Assign the free AI Fluency course as follow-up. The Claude Academy catalog lists it at about four hours, with AI Capabilities and Limitations as an optional second course.
• Teach one habit above all: give context and a standard, then check the answer against a source.

Week 3: Real tasks, with support#

This is where training becomes behavior:

• Each person applies AI to one of their three tasks, with a champion available.
• Save the instructions that worked as shared templates, or as skills if your team uses Claude. I explain those in Claude skills.
• For multi-step work like building a report from exports, try Claude Cowork, which handles files directly.
• Hold two short check-ins to share wins and failures. Failures teach Discernment faster than any slide.

Week 4: Measure, document, repeat#

• Re-time the tasks and compare with week 1.
• Keep what worked in a shared library with owners.
• Update the guardrails with what you learned.
• Record the program for compliance: dates, attendees, content, tools.
• Choose the next tasks for month two.

Different depth for different roles#

Not everyone needs the same program. I split it into four levels:

• Everyone: the guardrails, the 4D framework and one real task, about six hours across the month.
• Power users: people whose work is most AI-suitable get extra sessions on templates, skills and multi-step tasks.
• Champions: one or two per team learn to troubleshoot, collect what works and coach others once the 30 days end.
• Leaders: a short session on policy, risk and how to read the before-and-after numbers, so decisions rest on evidence rather than enthusiasm.

How to measure whether AI training worked#

Completion rates tell you almost nothing. Measure behavior and output instead:

• Time per task before and after, on the same tasks.
• Weekly active use of approved tools after 30 and 90 days.
• Reusable assets created: templates, skills or workflows other people use.
• Quality issues caught in review, which should drop as Discernment improves.

Common mistakes in AI training for employees#

• Generic demos. Writing a poem about the company doesn't transfer to closing the monthly books.
• Tool tours without tasks. People forget features they never needed.
• No guardrails. Without clear data rules, careful people won't use AI and careless ones will overshare.
• One-off sessions. Skills fade without week 3 practice.
• Leaders who don't participate. Teams copy what managers actually do.

Where to go next#

Start week 1 on Monday with the task list and the timing; it costs nothing and makes everything after it measurable. If you'd like this delivered for your team, remotely or in person in Madrid, see my team training. For technical teams ready to go further, how to build an AI agent is the natural next step, and Claude certification covers the formal credentials.

If the gap is wider than a workshop can close, an AI consultant for small business builds the first system with you and trains the team on it rather than on slides.