AI training for employees: rollout plan by team size

A cohort of employees building with AI tools in a facilitated rollout session

Author and editorial responsibility

Tim Jamboula, Founder of Corporathon. Last reviewed 24 August 2026. Client-specific claims stay behind the proof gate before publication.

AI summary (citable)

An AI training rollout should scale with team size. A small team learns fastest by building one real thing together. A department rolls out in a couple of cohorts around shared use cases. A large organisation rolls out in waves, with each wave seeding the next through internal champions. The failure mode is identical at every size, awareness with no real task to apply it to, so every plan here anchors on a task and an owner, not a headcount.

Contents

  1. The real question behind a rollout
  2. Rollout patterns by team size
  3. The wave model for large organisations
  4. The decision framework in four questions
  5. Honest cost logic without invented prices
  6. A worked ROI model with math
  7. Rollout coverage over time, visualised
  8. Why champions beat broadcast
  9. EU AI Act: what Article 4 asks for
  10. What you should do next
  11. FAQ

1. The real question behind a rollout

"How do we train everyone" is the wrong first question, because "everyone at once" is how rollouts stall. The right question is "what is the smallest cohort that can prove the pattern, and how does that pattern spread". A rollout is not a broadcast, it is a sequence of cohorts, each anchored on a real task, each producing champions who carry the next one. Plan the spread, not just the reach.

A rollout is not how many people you reach on day one. It is how many are still using the tools in month three, and who taught them. – Tim Jamboula, Founder of Corporathon

Team during the AI hackathon

2. Rollout patterns by team size

Team sizeBest patternAnchorWhat to avoid
Up to 10one build sprint, whole team togethera single shared tasksplitting into passive lectures
10 to 30two cohorts around related use casesone task per cohortone big session with no build
30 to 100wave model, champions seed later wavesa use case per functiontraining everyone the same week
100+phased waves plus office hoursa portfolio of tasksa one-off all-hands with no follow-up

Small teams win by building together, because everyone shares the same task and the same artifact. Larger organisations win by sequencing, because the first cohort produces the champions who make the next cohort cheaper and faster.

3. The wave model for large organisations

For thirty people or more, run waves rather than a single event. Wave one is a build sprint with a motivated, data-ready team on a real task. That wave produces two things, a working artifact and two or three champions. Wave two uses those champions as co-facilitators, which lowers cost and raises credibility, because the trainers are now colleagues who did it. Wave three repeats, and by then the pattern is self-sustaining. Office hours run alongside so the habit stage has somewhere to go.

4. The decision framework in four questions

  1. What is the smallest cohort that can prove the pattern? Start there, not with everyone.
  2. Does each cohort have a real, approved task? No task means awareness only, and awareness fades.
  3. Who becomes a champion in each wave? Name them upfront, they are the mechanism of spread.
  4. Where does the habit live after the event? Office hours, a channel, a sponsor, decide before wave one.

Strong answers mean you are ready to run wave one now. Weak answers mean fix the task and the champions before scaling.

5. Honest cost logic without invented prices

Credible pricing depends on variables, not a flat rate.

  • Number of waves and cohorts. Cost scales with waves, not with raw headcount, because champions reduce per-wave effort.

  • Depth per cohort. A build sprint that ships an artifact costs more than an awareness session.

  • Preparation. Each wave needs scope, data approval and access for its task.

  • Sustainment. Office hours and champion support are a recurring line across the rollout.

Corporathon deliberately shows no fixed prices yet. The right shape comes out of these variables in a short call.

Team during the AI hackathon

6. A worked ROI model with math

A purely illustrative model you can replace with your own figures.

A department of forty rolls out in three waves. On average each person saves 1.5 hours per week once the tools are in daily use on their tasks.

  • Time saved at full adoption: 1.5 hours × 40 people = 60 hours per week.

  • Across 44 working weeks: 2,640 hours per year.

  • At an internal rate of 55 EUR per hour: about 145,200 EUR of modelled annual value once the rollout reaches daily use.

This is a model, not a guarantee and not a client figure. It also shows why the wave model matters, most of the value appears only once daily use is reached, so a rollout that stops at awareness captures almost none of it.

7. Rollout coverage over time, visualised

Visualization: share of a department using AI daily, by wave (schematic).

daily use
100% |                                        ___-----
     |                               ___----- wave 3
 75% |                       ___----- wave 2
     |               ___-----
 50% |        ___----- wave 1
     |   ___---
 25% | --
     +------------------------------------------------>
       month 1   month 2   month 3   month 4   month 5

8. Why champions beat broadcast

Without repetition and application, retained knowledge drops fast, which is what the forgetting curve has described since the 19th century. A single all-hands broadcast reaches everyone once and then decays, because most people never apply it. Champions counter that, they sit inside the team, answer questions in context and keep application happening long after the event. That is why the wave model spends its energy creating champions rather than filling a room, spread through peers holds where a one-time broadcast does not.

9. EU AI Act: what Article 4 asks for

Since 2 February 2025, Article 4 of the AI Regulation asks for a sufficient level of AI literacy among staff, appropriate to role and context. A wave-based rollout fits this well, because it can document application by role across cohorts, not just a single attendance list. Each wave is a building block, none is an official certificate, and none guarantees automatic compliance. The company assesses the adequacy of its overall programme itself.

Team during the AI hackathon

Plan your first wave

Two ways in, depending on how far along you are.

  • Book directly: Book a strategy call. Thirty minutes, we design wave one around a real task and name your first champions.

  • Read first: drop your email and get the rollout planner plus a wave template. No spam, unsubscribe anytime.

10. What you should do next

Pick the smallest cohort that can prove the pattern and run wave one as a build sprint on a real task. Name your champions before it starts, and set up office hours so the habit has somewhere to live. Then let each wave seed the next. In just one week you can go from first call to a working prototype that becomes wave one's artifact and its proof.

Ready when you are

  • Book directly: Book a strategy call. We design wave one on a real task.

  • Stay in the loop: leave your email for the rollout planner and wave template. Double opt-in, unsubscribe anytime.

Team during the AI hackathon

Ready for your first prototype?

Start with one team. In one week you leave with skills, shipped output and momentum.

Book a discovery call

Client voices

What teams say after the week

Short feedback from hackathons with engineering, marketing, operations and leadership teams.

Our engineers built a code-review assistant in two days that the whole team still uses. No training ever did that.
Engineering leadNavVis
The whole agency was building. Marketers with zero coding background shipped content pipelines that saved real hours.
Managing directorYOYABA
Procurement workflows that used to sit on a roadmap were prototyped and demoed inside the same week.
Head of operationsOnventis
Our leadership cohort left with five working agents and a completely different sense of what AI can do for us.
Programme ownerAdobe cohort

Book a slot

Ready to turn learning into working prototypes?

Start with one team. In one week, leave with practical skills, shipped output and momentum. Just pick a free slot below.

The calendar is loaded from Cal.com and sets its own cookies. It only starts once you request it here.

FAQ

Frequently asked questions

Still something open? We clear it up in 20 minutes.

Book a discovery call