AI Upskilling Rollout: HR Plan for 30, 100 and 500 Employees

Tim Jamboula, Founder of Corporathon. Last reviewed on 25 August 2026.

An AI rollout rarely fails at booking courses, it fails at the missing transfer into real work. With 30 employees, one wave with an owner carries. With 100, you need multipliers and waves by function. With 500, you need a programme with internal champions, governance and a rhythm instead of a one-off session. Size does not change the goal, it changes the mechanics of spread.

"How do I upskill 500 people in AI" is the wrong first question. It leads to course catalogues, attendance lists and a good feeling in HR, but not necessarily to changed work. The right question is how trained people actually turn into teams that work differently, and how that carries through an organisation without fizzling out in daily work. A rollout is a spread problem, not a booking problem. That is why size changes the mechanics, not the goal.

An AI rollout is a contagion, not a memo. People adopt new tools when they see a colleague visibly getting faster with them, not because a mandatory training sat on the calendar.

- Tim Jamboula, Founder of Corporathon

The jump from 30 to 500 is not a volume problem, it is a structure problem. At 30, everyone knows everyone, one visible success is enough as a spark. At 500, every spark dies in the distance unless champions carry it forward and governance holds the frame.

Two paths, depending on how far along you are.

• Book directly: Book a discovery call. 30 minutes, we look at your size, structure and goals and sketch a rollout path with your numbers. To your Corporathon in just 1 week.
• Read along first: Sign up with your email and get the rollout planner plus real cases. No spam, unsubscribe anytime.

Answer these four questions honestly, and the rollout path is almost obvious.

1. 1How many people and how many functions? A homogeneous group of 30 carries one wave. Many functions need separate waves with their own case each.
2. 2Are there internal champions? Without multipliers who carry success forward, every session stays a one-off event. Naming champions is the first real rollout step.
3. 3Is it a session or a rhythm? From about 100 employees on, no one-off session carries. A quarterly rhythm anchors the topic before it fades again.
4. 4Who measures and is accountable? Without an owner for metrics and governance, the rollout becomes invisible and therefore unmanageable. That holds at 30 as much as at 500.

Rule of thumb, the bigger the organisation, the more the spread structure decides success, and the less the individual course does.

Credible prices depend on variables, not on a flat rate. Whoever quotes a number without these variables is guessing. For a rollout, these drivers weigh especially heavily.

• Number of waves, not heads. A rollout scales with the number of waves and functions, not linearly with headcount. 500 people in one homogeneous function are cheaper than 100 across five very different ones.
• Building the champion structure. Training multipliers costs once, but saves every further wave. This preparation is part of the cost and part of the value.
• Depth per wave. An awareness session per department is cheaper than a hackathon per department with a real prototype and handoff.
• Rhythm costs. A quarterly series has ongoing costs, but also ongoing impact. These follow-on costs honestly belong in the calculation.

Corporathon deliberately does not quote fixed prices on the website. The sensible shape emerges in conversation, from exactly these variables.

Costs alone say nothing without the benefit next to them. Here is a purely illustrative model, replace it with your own numbers.

A company rolls out AI to 100 employees. Instead of starting all 100 at once, three waves of roughly 33 people each run, each with a real case. Assume each person saves an average of 1.5 hours per week after the wave, through better tools and a built prototype.

• Time saved fully rolled out: 1.5 hours x 100 people = 150 hours per week.
• Over 45 working weeks: 6,750 hours per year.
• At an internal hourly rate of EUR 60: roughly EUR 405,000 in modelled annual value across all functions.
• Realistically weighed against that is an adoption rate, since not every person uses the new way equally. At an assumed 60 percent adoption, roughly EUR 243,000 remains.

This figure is neither a guarantee nor a customer figure, it shows the order of magnitude against which rollout costs can be checked. The decisive lever in the model is not the number of people trained, it is the adoption rate. That is exactly what a rollout with champions and rhythm works on.

New tools spread through organisations much like a contagion, through contact and visible success, not through directive. This observation has been known in diffusion research for decades, a small group of early adopters convinces the majority far more strongly than any mandatory training. A rollout that only books courses ignores this mechanism, and then is surprised that almost nobody uses what was learned after three months.

Champions are the early adopters who make success in their own team visible. Without them, the transmission path is missing, and even good courses fizzle out between waves, just as isolated knowledge decays along the forgetting curve without use. A hackathon works well as a starting point, because it immediately produces visible outcomes and therefore natural champions, whose prototype keeps running in daily work.

Since 2 February 2025, Article 4 of the AI Regulation requires sufficient AI literacy among employees, role- and context-specific. A rollout is exactly the frame in which this can be documented in a structured way, because it captures roles, functions and actual application. A single session documents little, a programme with waves, champions and metrics produces credible internal evidence. That remains one building block, not an official certificate, and does not guarantee automatic compliance. The company itself must assess the adequacy of the overall programme.

At 30 employees, one wave on a real case with a clear owner, then measure usage. At 100, a pilot in one department, win champions from it, then roll out by function. At 500, a flagship pilot, then a quarterly series with an internal champion structure and governance. At every size, start with a real case, not a course catalogue. Whoever is unsure clarifies the path fastest against their own structure.

• Book directly: Book a discovery call. We build a pilot with you that produces your first champions. To your Corporathon in just 1 week.
• Read along first: Sign up with your email and get the rollout planner. No spam, unsubscribe anytime.

AI Upskilling · AI Bootcamp · AI Driving Licence

• EU Regulation 2024/1689 (AI Act), EUR-Lex
• European Commission, questions and answers on AI literacy
• NIST AI Risk Management Framework