On this page
TL;DR:
- Decide in this order: what must change at work, who needs training, which format, what group size, then when and where.
- For most teams, groups of up to about 20 work well. Larger numbers do better split into several cohorts by type of work.
- Two or more shorter sessions with practice in between usually beat one long, tiring event, though the research has limits.
- In Nairobi, plan around traffic. Avoid late Friday finishes, set a realistic start time and let people travel against the worst of the jam.
- Train in the office when people need real systems. Go off-site when interruptions keep breaking the session.
Who this guide is for
This guide is for HR managers, learning and development leads, operations heads and founders who have been told to "get the team trained on AI". It covers the planning decisions that come before the first session: format, group size, timing and venue.
For the cost side, see our guide to what corporate AI training costs in Kenya. For choosing a trainer, see how to choose an AI training provider in Kenya.
Start with five decisions, in the right order
The right plan comes from the work your staff do, so the venue should be one of the last things you choose. It is tempting to book a date and a hotel first, then try to fill the day. Work through these questions instead.
| Order | Decision | The question to answer | Who should answer it |
|---|---|---|---|
| 1 | Outcome | Which three tasks should be faster or better 60 days after training? | Department heads |
| 2 | People | Who does those tasks, and what do they already know? | HR and line managers |
| 3 | Format | How much practice do they need, and how much time can be released? | HR and the trainer |
| 4 | Group size | How many cohorts does that create? | HR and the trainer |
| 5 | Timing and venue | When can each cohort attend, and where will they work best? | HR and operations |
A clear outcome sounds like this: "Finance can build the monthly variance commentary in half a day instead of a full day, with the same accuracy." A weak outcome sounds like this: "Staff should understand AI." You can only measure the first one.
Before you choose anything else, ask each department for its three most repetitive written or spreadsheet tasks. Our AI readiness assessment is another useful starting point.
Which training format should you choose?
Choose the shortest format that still gives everyone real practice on their own kind of work. A talk can raise interest. Skills come from doing the work with someone nearby to help.
| Format | Best for | Main risk | Our view |
|---|---|---|---|
| 90-minute leadership briefing | Executives who need to set direction and rules | People confuse awareness with skill | Useful before staff training starts |
| Half-day kick-off | Small teams with one clear use case | Too little time for mixed ability levels | A good pilot, as long as practice follows |
| One-day workshop | One department with shared tasks | Tired by mid-afternoon, and little follow-up | Works if a follow-up is already booked |
| Two-day workshop | Teams with Excel, reporting and writing needs | Two full days away from work | Strong if days are split by a week |
| Weekly sessions over several weeks | Habit building across a department | Attendance drops if managers do not protect the time | Often the best for lasting change |
| Live online sessions | Staff spread across Nairobi or the region | Cameras off and multitasking | Keep sessions short with tasks in every one |
| One-to-one | Senior leaders or a specialist role | Does not build a shared way of working | Good alongside team training |
Why spacing usually helps
Practice spread over time usually produces better performance than the same practice packed together. A well-known meta-analysis by Donovan and Radosevich, published in the Journal of Applied Psychology in 1999, combined 63 studies and found an overall effect in favour of spaced practice (mean weighted effect size 0.46).
The same paper has important limits. The size of the benefit depended on the type of task and the gap between sessions. Studies with weaker methods showed bigger effects. It is also an older review and was not about AI training.
So treat spacing as a sensible default rather than a guarantee. There is a practical reason too. A gap lets people try the skill on Tuesday's real report, get stuck, and bring the problem to the next session. Our piece on why corporate AI training does not stick goes further into this.
Avoid the hybrid room
Mixing people in the room with people on a video link often serves neither group. The trainer cannot see the online screens, and online staff miss the side conversations where much of the learning happens. If some staff cannot come, run a separate online session.
How many people should be in each group?
Keep groups to about 20 people so that everyone gets help during practice. That is how FuKazee plans its sessions. A trainer can only walk around so many screens, and in any new skill the slowest learners need attention first.
Group by the type of work people do. Seniority matters less, with one exception: some staff will not ask basic questions in front of their boss.
| Your situation | A sensible split |
|---|---|
| Under 20 people, similar work | One cohort |
| Under 20 people, very different roles | One cohort with role-based exercises in the afternoon |
| 21 to 40 people | Two cohorts, split by function |
| 41 to 80 people | Three or four cohorts, starting with one pilot group |
| More than 80 people | Start with a pilot and trained champions, then roll out in waves |
| Very senior leaders | A separate short session before the staff cohorts |
| Shift or branch teams | Run cohorts back to back so that some staff stay on duty |
A worked example. Suppose a company has 63 office staff: 18 in finance, 22 in sales and customer care, 12 in operations and 11 in HR and admin. A sensible plan has four groups, though HR and admin could join operations if the tasks overlap. Start with whichever team has the most willing manager. A strong first cohort gives you stories and prompts for the others.
Do not put everyone in the same room for "one big launch". A large audience can be inspiring for an hour. It is a poor way to teach Excel formulas.
When should you schedule training in Nairobi?
Choose the days around your work calendar first, then choose the hours around the roads. Most Nairobi teams should aim for Tuesday to Thursday, avoid their own peak periods, and finish well before the Friday evening rush.
The Nation's October 2025 mapping of Nairobi traffic names these daily gridlock roads: Thika Superhighway, Mombasa Road, Ngong Road, Jogoo Road, Lang'ata Road, Outer Ring Road, Uhuru Highway and Waiyaki Way. It also reports that jams are no longer limited to rush hours and can hit mid-morning or early afternoon. A driver quoted there called Friday evenings the worst, saying the city locks up by 4pm. Treat that as one person's experience, though many Nairobi workers will recognise it.
Your work calendar comes first
Block out these periods before you look at dates:
- Month-end and payroll for finance and HR staff.
- Quarter-end, audits and board papers.
- Your own sales peaks and busy customer periods.
- Any major system migration or office move.
- Periods when many staff are on leave.
Then plan around the roads
| Decision | Practical advice for Nairobi |
|---|---|
| Start time | Ask where people begin their day. Offer tea from 8:30 am and start teaching at 9:00 am. Teams that travel against the main flow can start earlier |
| Early starts | Useful for shift teams. We have run 8:00 am starts for a client in Mombasa |
| Finish time | Aim for 4:00 pm on most days and earlier on Fridays |
| Monday mornings | Often taken up by weekly meetings and emails |
| Fridays | Fine for a half-day, and poor for a full day |
| Multi-site companies | Choose a central site for the cohort, or go online if people must cross the whole city |
| Arrivals | Allow a 30-minute buffer at the start, with a useful task for early arrivals |
| Rain | Tell people not to rush, and have a recap ready for the first half hour |
If most of your staff use one route, look at that road specifically. Teams coming to a venue along Waiyaki Way, Mombasa Road or Thika Road may need a very different start time from a team based within walking distance.
Saturday training can look tempting because the roads are quieter. Our view is that it can tell staff that learning is an extra burden, so use it only when staff have truly agreed and the company treats it as work. Check your employment terms first.
Should you train in the office or off-site?
Train in the office when staff need their own systems and real files. Go off-site when the office would keep interrupting the session. Score your own situation with this table.
| Factor | In your office | A venue in Nairobi | A retreat outside Nairobi |
|---|---|---|---|
| Access to real systems | Strong | Depends on your IT setup | Weak to medium |
| Freedom from interruptions | Weak unless managed | Strong | Very strong |
| Travel time | Lowest for most staff | Medium | Whole-day impact |
| Practice on real work | Strong | Medium | Medium, with masked files |
| Team bonding | Limited | Some | Strong |
| Internet reliability | Known | Must be tested | Must be tested |
| Extra costs | Lowest | Venue and meals | Venue, meals, transport and possibly rooms |
In the office works if you set clear rules. Use a closed room, away from desks. Tell clients and colleagues that the group is unavailable. Managers must leave staff alone, including on WhatsApp.
A Nairobi venue suits a team that keeps getting pulled back to its desks. Westlands is one option, and we explain how training works at our base on the AI training in Westlands page. Waiyaki Way is on the Nation's congestion list, so people coming from far away should allow enough time.
A retreat suits leadership teams making decisions. See AI training in Naivasha if you are thinking of going out of town. Remember that a retreat costs travel time on both ends.
What must be ready before training day?
Most lost training time comes from small setup problems: forgotten passwords, blocked websites and no sample files. Go through this list one week before.
Accounts and access
- Each person has a working laptop and knows their email password.
- Multi-factor sign-in works, and people have their phones.
- ICT has confirmed which AI tools are approved.
- The network does not block the sites you plan to use.
- If you use Microsoft 365, you know exactly which staff have which Copilot licences.
Content
- Each person has named two or three tasks they want to improve.
- Sample files are ready, with names, phone numbers, ID numbers, M-PESA details and salaries masked.
- The first version of your AI rules exists, even if it is short. Our guide to an AI usage policy for Kenyan organisations can help.
The room
- A power point within reach of every seat.
- A screen that can be read from the back.
- Wi-Fi tested with the whole group's devices in mind.
- A backup hotspot and printed exercises.
- Tables arranged so that a trainer can reach every person.
Masking matters even during training. Kenya's Data Protection Act treats details such as health status and family details as sensitive personal data. Practising on real customer or staff records in a public tool creates exactly the habit you are trying to prevent.
Tool settings matter as well. Microsoft says prompts in Copilot Chat, as used by organisations, are covered by enterprise terms and are not used to train its foundation models. Google's Gemini privacy page warns that a subset of chats may be seen by human reviewers. Check each tool's current terms before staff use it for work.
A sample training day
This shape works for a one-day session with up to about 20 people. Adjust it to your team.
| Time | What happens | Why |
|---|---|---|
| 8:30 to 9:00 | Tea, logins and a short warm-up task | Absorbs late arrivals and sign-in problems |
| 9:00 to 10:00 | What AI can and cannot do, and safe use | Shared rules from the start |
| 10:00 to 11:00 | The four Cs: Clarity, Context, Constraints, Customisation | A method for every tool |
| 11:00 to 11:20 | Break | Time to answer messages, so people do not do it later |
| 11:20 to 1:00 | Practice on masked real work, by function | Where most learning happens |
| 1:00 to 2:00 | Lunch | Protect it |
| 2:00 to 3:15 | Excel, Word or reporting exercises | Applied work helps in the sleepy hour |
| 3:15 to 3:45 | Checking AI output and catching errors | The habit that prevents embarrassment |
| 3:45 to 4:00 | Each person writes one task to try this week | Links the training to real work |
Three prompts for the person planning the training
These prompts help you plan. Remove names before you paste anything.
1. Find the themes in your staff survey
Below are anonymous answers from staff to the question "Which two tasks take too much of your time?" Group them into no more than six themes. Context: we are planning AI training for a Nairobi company. Constraints: show a table with the theme, the number of answers, one short example and the department most affected. Do not add tasks that are not in the answers. Customisation: the reader is our HR director, who will use this to choose cohorts. Answers: [paste with names removed]
2. Build a cohort calendar
Help me schedule AI training cohorts. Context: we are based in [area] of Nairobi. Cohorts: [list with group sizes]. Blocked periods: [list]. Constraints: use Tuesday to Thursday where possible, keep each group to 20 people or fewer, leave at least one week between Session 1 and Session 2, and never run finance during month-end. Customisation: give a table with cohort, dates, start and finish times, and any clash you could not resolve. Mark anything you are unsure about with [CHECK].
3. Write the invitation staff will actually read
Write an email from our HR manager inviting [team] to AI training on [dates] at [venue]. Context: some staff are worried about losing their jobs, and many have never used an AI tool. Constraints: under 150 words, friendly, plain English. Say that no experience is needed, what to bring, what time to arrive, and to bring two tasks they would like to improve. Do not promise anything about jobs. Customisation: this will be read quickly on a phone.
What goes wrong, and how to avoid it
| What goes wrong | Why it happens | What to do |
|---|---|---|
| Half the group arrives late | Start time ignored the roads | Ask where people start their day, and use a 30-minute buffer |
| The first hour is lost to logins | Nobody checked accounts | Run an access check a week before |
| Managers keep pulling people out | Time was never protected | The sponsor tells managers in writing |
| Beginners feel lost and experts feel bored | One group, mixed tasks | Group by work, and give stretch tasks |
| Someone pastes real customer data | No rules or masked files | Written rules and prepared examples |
| The Wi-Fi fails | Venue not tested | Test in the room, and bring a hotspot |
| Nothing changes after the day | No follow-up was planned | Book the 30-day check before the training |
| Staff fear for their jobs | Leaders said nothing | Leaders explain the purpose honestly at the start |
After training: the first 30 days
Training only pays off if people keep using it. Plan these steps before the first session.
- Week 1: Each person tries one task and notes what happened.
- Week 2: The team shares its best prompts in one shared document.
- Week 3: A manager reviews one real AI-assisted piece of work with each person.
- Week 4: Hold a one-hour follow-up. Compare time taken, including checking and correcting.
Nominate one or two champions for each department. Choose careful people, not only the most excited ones. Give them a little protected time, or the role will quietly disappear. Our guide to getting middle managers to use AI explains why line managers decide whether any of this lasts.
Questions people ask
Is a half-day enough to train staff on AI?
It can be enough for a first step with one clear use case. It will not make anyone an expert. If you have only half a day, cover safe use and one practical task, then book a follow-up.
How many staff can be trained at once?
We plan around groups of up to about 20, so that everyone gets help during practice. Larger organisations can run several cohorts. A big single session suits awareness better than skill-building.
Is it better to train staff at the office or at a hotel?
Your office is usually better for hands-on work with real systems, provided you can stop interruptions. A hotel or other venue helps when staff keep being pulled away. Test the internet in any venue first.
What is the best time to run AI training in Nairobi?
Tuesday to Thursday usually works best, starting at about 9:00 am with an early buffer. Avoid month-end for finance teams and long Friday sessions. Adjust the times according to the routes your staff use.
Should senior managers attend with their teams?
Give senior leaders their own short session first. They need to set direction, approve tools and agree the rules. They can then join part of the staff session to show support.
Can staff across Nairobi learn online?
Yes. Live online training saves travel and suits staff spread across the city. Keep sessions short and give each person a task, because attention drops quickly on a video call.
Do staff need paid AI tools?
They can start with free versions for learning. If your company uses Microsoft 365, include Copilot and check who has which licence. Real work should go only into tools your company has approved.
Sources
- Daily Nation, "Gridlocked capital: Mapping Nairobi's worst traffic hotspots and timings" (27 October 2025), last checked 9 November 2026
- Donovan and Radosevich, "A meta-analytic review of the distribution of practice effect", Journal of Applied Psychology, 1999, last checked 9 November 2026
- Data Protection Act, 2019, copy on the Data Commissioner's website, last checked 9 November 2026
- Microsoft, "Enterprise data protection in Microsoft Copilot and Microsoft Copilot Chat", last checked 9 November 2026
- Google, Gemini Apps Privacy Hub, last checked 9 November 2026
Where FuKazee fits in
FuKazee runs AI training from Westlands, Nairobi. We can come to your office, work at a venue you choose, or host you at our base at 124 Manyani East Road, near ABC Place. Since 2022 we have trained more than 3,000 people.
We assume nobody knows AI, we avoid jargon, and we say plainly that nobody becomes an expert in a few days. We show the free version of every feature and never promote one tool, and Microsoft 365 and Copilot are part of what we teach. Tell us your people, your calendar and your three most painful tasks. We will suggest a format that fits, and we will say so if a shorter programme would do.
See AI training in Nairobi, AI training in Westlands, or our wider page on AI training in Kenya.
To plan training for your team, email us at 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.
Rodney has run FuKazee's training since 2022 and still teaches inside every programme line: corporate sessions, the postgraduate cohort and the KCSE examiner webinars. First Class Honours in Journalism and Mass Communication, University of Nairobi.
