On this page
Kenyan companies that train large teams on AI usually use one or more of five approaches: live groups by role, an internal champions network, self-paced e-learning with some live sessions, programmes from the software vendor, or an external trainer who runs the kick-off. What tends to work best is a mix: a short live start on real work, managers first, clear data rules and a few champions to keep it going.
TL;DR:
- No single approach works alone. Most successful programmes combine live training with something that lasts after it.
- Live groups of about 20, by type of work, teach the most. E-learning is cheap but easy to skip.
- Champions keep AI use alive, but only with protected time.
- Set the data rule before anyone practises. Kenya's Data Protection Act, 2019 applies.
- Measure real tasks, not logins.
Why large companies are training now
Staff are already using AI, with or without training. Microsoft and LinkedIn's 2024 Work Trend Index, a survey of 31,000 people in 31 countries published in May 2024, found that 75% of knowledge workers used AI at work. It also found that 78% of AI users brought their own AI tools to work, and only 39% of people who used AI at work had received AI training from their company.
That survey is global, not Kenyan. But the pattern is familiar: people experiment alone, with no shared method and no rule on what data stays out. Training a large team is how a company turns that into safe, consistent use.
The five common approaches
| Approach | Good for | Watch out for |
|---|---|---|
| Live groups by role | Real practice on real work; questions answered | Takes staff time; needs careful scheduling |
| Champions network | Keeping use going after training | Burns out champions without protected time |
| E-learning plus live sessions | Reaching many people cheaply; new joiners | Low completion; nobody checks the work |
| Vendor programmes | Learning one product in depth | Centred on that product |
| External kick-off, then in-house | A fast, shared start, then lasting support | Needs champions ready before the trainer leaves |
1. Live groups by role
Staff learn in groups of about 20, sorted by function: finance with finance, sales with sales. Each group practises on its own documents with masked data. This is slower to organise but it is where people actually change how they work. For how this plays out for a company of about 200, read how AI training for a 200-person team works.
2. A champions network
A few people in each department who already use AI well become the first place colleagues go for help. Microsoft runs a Microsoft 365 Champion community and publishes a guide to starting your own champion community. The same idea works for any AI tool. Our guide on running an AI champions programme covers choosing and supporting them.
3. E-learning plus live sessions
Some companies assign free online courses, then add a few live sessions. Microsoft says its Microsoft Learn training is free. Elements of AI, from MinnaLearn and the University of Helsinki, describes its Introduction to AI as a free online course with "no complicated math or programming required". Our free AI courses in Kenya page lists more. E-learning teaches the basics well. It cannot see a person's own report or spreadsheet.
4. Vendor programmes
Software companies offer training on their own tools. This helps if you already pay for that product. Expect their product in every example, and keep a neutral view when you decide what to buy next.
5. External kick-off, then in-house
An outside trainer gets everyone to the same starting point within a few weeks, and trains champions who then run refreshers and onboard new staff. Our page on in-house vs external AI training compares the two in detail.
What works
- Start with tasks, not tools. Ask each department which reports, letters or analysis take the most time.
- Train managers first. They decide whether staff get time to practise.
- Keep groups to about 20. Larger rooms leave quiet people behind.
- Set the data rule first. Staff and customer records fall under Kenya's Data Protection Act, 2019, enforced by the Office of the Data Protection Commissioner. Real names, ID numbers and M-PESA statements stay out of public tools. A short AI usage policy helps.
- Practise on real work. Generic examples are forgotten by Monday.
- Protect follow-up time. Nobody becomes an expert in a few days. Short refreshers after 30 and 90 days help habits stick.
- Measure tasks. Time a few real tasks before and after, including the time spent checking.
Training is one part of a wider programme that also covers pilots, licences and policy. For the full picture, read our AI rollout plan for companies with 100+ staff.
Common pitfalls
- One big town hall for everyone. It raises interest, then nothing changes.
- Buying licences before training. Seats go unused when nobody knows what to use them for.
- The same content for every team. Drivers and finance analysts need different examples.
- Training, then silence. Without champions or follow-up, use drops within weeks.
- Counting logins. A login does not mean the work improved.
Kenyan realities to plan for
- Branches across counties. Live online groups save travel, if people have a laptop and stable internet. Our post on online, on-site or hybrid AI training weighs the options.
- Shifts and month-end. Stagger groups so counters and desks stay covered, and keep finance free at month-end.
- Kiswahili outputs. Anything AI writes in Kiswahili for customers should be read by a fluent speaker first.
- The training levy. NITA's levy page says employers pay the Industrial Training Levy at KSh 50 per employee per month, and that levy-compliant employers can have staff training costs taken care of by NITA "based on Levy terms and conditions". The same page asks employers to seek NITA's approval before the training. Check those terms, including which courses and providers qualify, before you plan around a refund.
Questions people ask
What is the cheapest way to train a large team on AI?
Free e-learning costs the least in fees. It costs more in staff time if few people finish. Many companies pair it with a short live session per group.
How long does it take to train hundreds of staff?
Usually weeks or months, because groups run in sequence around operations. A 200-person team might need about ten to twelve groups.
Should IT run AI training?
IT should own the tools and security rules. The training itself works better when managers and champions from each department take part.
Can we just tell staff to take an online course?
You can, but completion is often low and nobody checks whether work improved. Add a live session or champion support.
Do we need an AI policy before training?
You need at least a short rule on what data stays out of AI tools. A full policy can follow, shaped by what staff learn.
Sources
All sources last checked 14 November 2026.
- Microsoft WorkLab, "AI at Work Is Here. Now Comes the Hard Part" (2024 Work Trend Index, 8 May 2024): 31,000 people in 31 countries; 75%, 78% and 39% figures.
- Microsoft Adoption, Become a Champion: Microsoft 365 Champion community and guide.
- Microsoft Learn, FAQ: Learn training is free.
- Elements of AI: free courses by MinnaLearn and the University of Helsinki.
- NITA, Levy Inspectorate: KSh 50 per employee per month; training costs under levy terms; approval before training.
- Office of the Data Protection Commissioner: mandate under the Data Protection Act, 2019.
- FuKazee, AI training in Nairobi: groups of up to about 20.
Where FuKazee fits in
FuKazee trains Kenyan teams on AI in live groups built on their own work, with managers first and champions left behind. We are not yet NITA accredited, so ask us for the current status if the levy matters to you.
See how we work on our AI training in Kenya page, or tell us about your team. Email lead@fukazee.com
Want guided support instead?
Plan AI training around your team's real work. Answer a few quick questions about your sector, team size and goals, and we will recommend the right programme and format. Sessions are built on your workflows and include a usage policy.





