Team competence
Put AI to work, safely.
Most companies already have AI tools; few people were taught what to use them for and what not to. Cyberact teaches that in about two minutes a day, and along the way keeps company data out of the wrong tools and people out of AI-written fraud.
Price on a 12-month subscription, excl. VAT. No per-seat fees. Renews automatically until you cancel. 14 days free on a 12-month subscription – card required, no charge until the trial ends.
What the research measured
Aimed at the right work, AI makes people measurably faster.
+14%
issues resolved per hour by 5,179 customer-support agents given an AI assistant, and +34% for the least experienced
Source: Brynjolfsson, Li & Raymond, Quarterly Journal of Economics 2025
−40%
time taken on professional writing tasks, with 18% higher quality, by 453 professionals given ChatGPT
Source: Noy & Zhang, Science 2023
+25%
speed on 18 realistic consulting tasks within AI's reach, with 12.2% more tasks done and better quality, among 758 consultants
These are single tasks in controlled studies, not company growth. That is why the calculator below starts from a far smaller number.
But
Aimed at the wrong work, the same tool makes results worse.
In the same consulting study, on a task chosen to lie beyond what AI could do, consultants using AI were 19% less likely to reach the right answer than those working without it. In another trial, experienced software developers took 19% longer with AI, and afterwards believed it had made them 20% faster. The difference isn't the tool. It is whether people know where the line runs.
Source: Dell'Acqua et al., Harvard Business School 2023 · METR, 10 July 2025
The gap
The licences are bought. The know-how isn't.
20% of EU enterprises with ten or more staff used AI in 2025, up from 13.5% a year earlier, and 55% of large ones. Among enterprises that had considered AI but not used it, the most common reason was a lack of relevant expertise, named by 71%.
78% of people who use AI at work bring their own tools, and only 39% have had AI training from their employer.
Five hours of training or more made regular AI use significantly more likely. A clear plan raised measurable business impact by 25 points; better tools alone, by 5.
Source: BCG, AI at Work 2025 and 2026
Training is the one variable a company controls.
What your people learn
Worth More: 41 lessons on turning hours into results.
Plus Inside the Machine, 37 lessons on how language models work and where they fail.
- 01
Where the hours actually are
- 02
Getting output you can send
- 03
The everyday jobs
- 04
From your habit to the team's
- 05
Automating a process
- 06
Turning hours into business
- 07
Doing it safely
- 08
The AI Act in practice
Safely, at the same time
A ban doesn't stop AI use. It moves it out of sight.
More than half of employees would use AI tools even if they were not authorised to.
Source: BCG, AI at Work, 26 June 2025
A high level of unapproved AI use added USD 670,000 to the average cost of a breach, and 63% of the breached organisations studied had no AI governance policy or were still developing one.
Source: IBM, Cost of a Data Breach 2025
Cyberact teaches what may go into an AI tool and what may not, how to spot AI-written fraud, and what the AI Act requires. Every completion is recorded as an Article 4 measure.
Work it out
The price is fixed. The return grows with every person who learns to use AI well.
Arithmetic on your own assumptions. Nothing you enter leaves this page.
- Time freed, a year
- €8,208
- Price, a year
- − €1,068
- People's learning time (1.4 h each)
- − €2,660
Pays back when 3 of 50 people become regular AI users because of the training.
Defaults: time saved is half the 5.4% of work hours that generative-AI users report themselves (St. Louis Fed, 27 February 2025), because self-estimates run high. Labour cost is the euro-area average for 2025, €38.2, rounded down (Eurostat). The extra users are an assumption: change them.
Source: St. Louis Fed, 27 February 2025 · Eurostat, 31 March 2026
Arithmetic on your assumptions, not a promise. Time saved isn't money until it is spent on something, which is exactly what the path's module on turning hours into business covers.
Price on a 12-month subscription, excl. VAT. No per-seat fees. Renews automatically until you cancel. 14 days free on a 12-month subscription – card required, no charge until the trial ends.
The competency model
Six competencies, measured separately.
Every lesson and every question is mapped to the competency it develops. That is why you can see that engineering is strong technically and weak on fraud — and point the training there rather than at everyone.
Security awareness
Recognising and resisting everyday attacks
Fraud resilience
Payment, invoice and impersonation fraud
Data protection
Handling personal data lawfully
AI literacy
Using and overseeing AI responsibly
Incident response
Spotting, reporting and containing
Technical depth
Encryption, access control and secure development — the skills that pay more
AI literacy is the one that compounds
The same skill repeats in every task AI suits, so its effect compounds. It doesn't show in a training report but in lead times and quality, which is why the calculator above lets you estimate it yourself.
Why it works
Two minutes a day beats two hours a year.
It fits in a working day
A two-minute lesson needs no scheduling and no conversation with a manager. Which is why it also gets done.
Review handles the forgetting
The question comes back just before it would have gone. An annual course does not do this, which is why its effect reads as zero in February.
Difficulty follows the person
The capable do not sit through foundations and the beginner does not drown. One programme serves both without anyone building a separate path.
It feels voluntary
Streaks, levels and optional leaderboards. No lives, no timers, no pressure — participation comes from it being interesting.
And it accumulates into the record on its own
Developing competence and proving compliance are not two projects. The same completion that raises a person's level is filed against the requirement it satisfies. Nobody assembles a folder separately for the audit.
See the manager consoleThe questions a business leader asks
We already bought Copilot or ChatGPT.
Then the licence is paid for, and the gap is use: only 39% of people who use AI at work have had AI training from their employer. Training is how the licence starts earning its keep.
People won't change their habits.
That is why it takes two minutes a day rather than a workshop, and why the path starts from each person's own week rather than from features. In one experiment, professionals who had tried ChatGPT were twice as likely to report using it in their real job two weeks later.
AI is a risk, so we ban it.
More than half of employees would use AI tools without permission, and frequent use of AI at work tripled in a year. A ban moves the risk out of sight; teaching the rules is cheaper than finding the leak.
We can't measure it.
Measure what can be measured: competence and retention by department, in the console. For time, use the calculator with your own figures, and compare one team's lead times before and after. We don't claim the report measures productivity.
Isn't the productivity story hype?
Across a whole workforce the effect is modest so far: in one US survey, the time saved came to 1.4% of all work hours, counting those who don't use AI. That's why the calculator starts from half of what users report. The big gains come on the right tasks, the losses on the wrong ones, and telling them apart is what the training teaches.
We have no time for training.
The 41 lessons of the AI path take about two minutes each: under an hour and a half per person, spread over weeks. The calculator counts that time as a cost.
Our experienced people don't need it.
The gain is largest for the less experienced, so for the basics that is fair. But the trial in which experienced developers were 19% slower while believing they were faster was about experts. Difficulty adapts, so the capable skip the foundations.
For business
For your security lead, it is also a legal duty
The same completion counts as evidence for Article 4 of the AI Act and for the training measure in your country's NIS2 law (NIS2 Article 21(2)(g)). There is no need for two projects.
The security lead's viewStart with everyone, for the price of starting with ten.
A flat price with unlimited users, a 14-day free trial on a 12-month subscription for new customers, and you can cancel any time. The training without the manager console is Platform, at €49/mo.
Price on a 12-month subscription, excl. VAT. No per-seat fees. Renews automatically until you cancel. 14 days free on a 12-month subscription – card required, no charge until the trial ends.