AI Literacy Training
Most teams get a company AI account and no idea what to do with it beyond writing emails. I train your people on the work they actually do: what these tools are good at, what they quietly get wrong, and what must never be pasted into them. Sessions run on your own documents and workflows.
AI literacy training teaches employees what AI tools can and cannot do, how to use them productively on real work, and where the risk lives: data that must never reach a public model, output the model invented, and confident answers nobody checked. Article 4 of the EU AI Act makes a sufficient level of AI literacy mandatory for companies deploying AI systems in the EU. Oleg Sotnikov runs the training himself, using the same material he presents at Google HQ and the workflows that took his own company to an AI-first team of two.
What the Training Covers
AI training for employees only works on their real tasks, so every track below runs on your workflows rather than canned exercises.
Tool skills on real work
We work on your documents, your tickets, your code, not a sample dataset. Everyone leaves having finished a task they were going to have to do anyway.
Prompt craft that holds up
Prompting that survives contact with a forty-page contract and a messy spreadsheet: context, constraints, examples, and how to describe what a good answer looks like before the model guesses.
Risk literacy
What never goes into a public model, how to check an answer that sounds right, and when to stop trusting a tool. Most AI incidents start with someone pasting the wrong thing into a chat box.
Role-specific tracks
Engineering, operations, sales and support, leadership. An engineer reviewing generated code and a support lead drafting replies need different hours and different examples.
EU AI Act Article 4 alignment
Training scoped, delivered, and recorded so your compliance file has something concrete to point at: who attended, what was covered, when it gets refreshed.
Before and after measurement
A short skill assessment at both ends. You see who moved, who needs a second session, and whether the training changed anything at all.
How It Runs
Baseline and tracks
A short assessment plus a conversation with team leads: what people already use, what they avoid, what the company needs them to be able to do. That decides the role tracks and the hours each one gets.
Sessions on your own work
Live sessions built around your tools and workflows. People bring current tasks and we do them together, including the ones where the model gets it wrong and someone has to notice.
Records and a refresher path
You get the attendance and coverage record for the compliance file, the material to onboard new hires with, and a refresher cadence so the training does not quietly expire.
Why Me
- I present at Google HQ on applied AI, and the training is built on that material
- I took AppMaster from 25 people to an AI-first team of 2, now running 11B+ tokens a month in production, so the material comes out of my own company rather than a case study
- 25+ years teaching engineers and executives, and 45+ certifications of my own, so I have a working sense of what learning something new actually costs someone
Article 4 of the EU AI Act
The training is worth doing on its own terms. The regulation attached a date to it.
What the article says
Providers and deployers of AI systems must ensure a sufficient level of AI literacy among their staff and the other people who operate those systems on their behalf. The obligation has applied since 2 February 2025.
Who it applies to
Companies that deploy AI systems in the EU, not only the ones building models. Enforcement and penalty provisions apply from August 2026, which is why the question is landing on desks now. Outside the EU nothing obliges you, and the productivity argument stands on its own.
What sufficient means in practice
The text prescribes no curriculum. In practice it means training matched to the roles involved and the risk of the systems in use, documented well enough that you can show what you did and when.
Frequently Asked Questions
What is AI literacy training?
It teaches employees to use AI tools on their own work and to recognize where those tools fail. The practical half covers prompting, checking output, and choosing the right tool for a task; the other half covers risk — what data must never leave the company, how to spot an invented fact or citation, when a confident answer deserves a second reader. It is aimed at everyone who touches the tools, not only engineers.
Is AI literacy training mandatory?
In the EU it is, for many companies. Article 4 of the EU AI Act requires providers and deployers of AI systems to ensure a sufficient level of AI literacy among the staff who operate them; that obligation has applied since 2 February 2025, and the enforcement and penalty provisions apply from August 2026. Outside the EU no law requires it, and the productivity case stands alone.
What does the training cover?
Practical tool use on your real tasks, prompt craft that holds up against long documents and messy data, and risk literacy: data handling, hallucination checking, and when not to trust output. On top of that sit role-specific tracks for engineering, operations, sales and support, and leadership. Every company gets a different mix, and the assessment at the start decides it.
How long does AI literacy training take?
It is scoped to the company. One team with one tool needs a fraction of what a hundred-person organization with four role tracks and a compliance file needs. The baseline assessment sets the hours before anyone commits, and the sessions usually spread over a few weeks rather than run as one long day, because people need time to use what they learned between them.
How is this different from an AI policy?
A policy sets the rules: which tools are approved, what data may go where, who signs off on what. Training builds the ability to follow those rules and the judgment for the cases the policy never anticipated. The two reference each other, and most companies want both. Write the policy first if nothing is written down; start with the training if the rules exist and nobody follows them.
The Pieces Around the Training
A policy says what is allowed; training is what lets people follow it. These cover the rest.
Train the Team That Already Has the Tools
A short call to scope it: how many people, which roles, what they use today, and whether the EU AI Act applies to you.
Free 30 minutes. Training is scoped per engagement.
Related reading
AI adoption, team capability, and the governance around both.
