Why Do Organisations Keep Choosing FuKazee for AI Training?

TL;DR:

• Three things recur in what FuKazee's clients have written: training built around their real work, plain language for non-technical people, and care with sensitive information.
• The evidence has limits. We publish letters from three organisations. Only one describes an employer training its own staff, and one concerns curriculum design, not a training room. None measures what changed months later.
• As of 8 November 2026, our site shows a Google rating of 4.9 from 38 reviews. Most of the reviews displayed there come from individual learners, so they say more about how learners felt than about results for employers.
• Before you buy from anyone, including us, speak to the person who signed the letter. Ask what staff were still doing three months later, and what went wrong.
• FuKazee does not sell software and does not promise results.

Who this is for, and how to read it

This piece is for HR managers, learning and development leads, procurement officers and directors who are comparing AI training providers and want to know how far to trust what they are shown.

FuKazee wrote it about FuKazee, so the bias is obvious. To keep it checkable, we linked each public page we used. Where a claim rests on our own word, we say so. The buyer's sections further down work for any provider in any sector.

What do FuKazee's clients say they valued?

Three things come up repeatedly in the letters and reviews on our site: training built around the client's work, plain language, and care with sensitive information.

Training built around their work. The letter from Maisha Girls Safe House calls the training practical and tailored to the organisation, and records a discovery session held before it. One Google reviewer describes being sent a discovery document at the start, to pin down what they needed.

Plain language. HackHouse Africa's letter says the session avoided unnecessary technical jargon and left founders with steps they could use. AfricaHackon's letter praises how the curriculum we designed simplifies complex AI concepts for people of any technical background. One Google reviewer, who says they barely knew computers, describes understanding the concepts anyway.

Care with sensitive information. Abraham Liona, Finance and Administration Manager at Maisha Girls Safe House, wrote that staff learned to use AI safely, "with careful attention to protecting confidential information". The organisation provides shelter and care for girls who are survivors of sexual violence. Our account of that training sets out what this meant in practice. The rules on what never goes into an AI tool came on the first morning, before anyone opened ChatGPT. Every exercise used invented or sanitised material.

Two limits apply to all of this.

Letters record what a client valued afterwards. None of ours says why the client picked FuKazee over another provider, so the public record can only partly answer the question in our headline.

The word "keep" also needs care. HackHouse Africa says it would be glad to work with us again and AfricaHackon that it would be glad to continue, which are statements of willingness. Our public pages do not document a repeat purchase. They do show a list of more than 20 organisations and institutions, and that count is ours.

What did each client actually buy?

Each of the three letters describes a different kind of work. Only one describes an employer training its own staff, which is what most HR managers are buying.

Maisha Girls Safe House

This is the closest match to a normal staff training purchase. Leadership, finance, monitoring and evaluation, communications, psychosocial support, programme staff and interns trained in the same room. The second day was built around each person's own work.

The letter lists a training report, an AI Playbook with prompt templates for each role, and certificates. Our account adds a participant guide and 14 days of follow-up support.

The letter is dated 12 days after the training ended. It is good evidence of how the training was delivered. It cannot tell you what staff were doing three months later.

HackHouse Africa

HackHouse invited FuKazee to train founders. They were not its employees, so this was community training and differs from a staff programme. The recommendation is explicitly for training a non-technical audience. If your staff are technical, it tells you less.

Two facts should affect how much weight you give this letter. FuKazee's office is at HackHouse Africa, so the two organisations know each other beyond one session. And the signer, Innocent Magothe, is also the founder of Startinev, a separate organisation that appears among the logos on our site.

AfricaHackon

AfricaHackon commissioned a curriculum. We wrote a base version, built in places for AfricaHackon to add its cybersecurity expertise, and revised it with the organisation's feedback.

This was curriculum design only. The letter praises how we listened to AfricaHackon's requirements, incorporated its feedback throughout development and simplified complex AI concepts without losing their practical value. That is evidence about how we design a programme and take feedback, and it tells you nothing about a training room. FuKazee did not teach it, and we claim no learner results from it.

The other names on our site

Our site also names organisations including Savannah Tracking, A.O. Bayusuf, Vicentia East Africa, the University of Nairobi and the US Embassy in Kenya. No letter from these organisations is published there, so for now the evidence is our word until you check it. That says nothing about how the work went. A logo on a provider's site, ours included, is also no endorsement by that organisation.

How much should you trust a 4.9 Google rating?

Treat it as a sign of how learners felt, and check the current figure on Google yourself. As of 8 November 2026, our reviews page, our About page and our vendor page show 4.9 from 38 Google reviews. When we tried to read the figure from Google's own listing, the page returned no rating or count, so 38 is what our site states. Our reviews page links to the listing, and the number will move.

Keep these points in mind.

• The sample is small. Thirty-eight reviews is a modest number. One or two people can shift the average.
• What a provider displays is chosen. Our reviews page says its reviews are a selection, shortened to the key passage. Read the full set on Google, starting with the lowest-rated and the newest.
• Most reviewers are individuals. Of the 23 reviews on our page, by our count only three speak for a team or an organisation. The rest come from individual learners, such as students, researchers, freelancers and business owners, speaking for themselves. One of the three is from Maisha Girls Safe House, so count its letter and its review as one voice.
• Superlatives tell you little. Several reviews call us the best. No reviewer has compared every provider in Kenya, so discount that word wherever you see it, on our page too. Look for reviews that describe what was taught, what the reviewer now does differently and what happened after the course.
• Platform rules help, up to a point. Google's policy says fake engagement is not allowed, and it bars businesses from offering incentives in exchange for reviews. A policy cannot prove that every review was earned, so look for detail you could only know from being there.

What should buyers look for in this kind of evidence?

Look for specific claims tied to a named person you can reach, and rank each piece of proof by how hard it would be to fake.

The Kirkpatrick Model, a long-standing framework for evaluating training, gives a useful way to sort it. It has four levels: reaction, learning, behaviour and results. Reaction asks whether people found the training engaging and relevant. Results asks whether the organisation's targeted outcomes happened because of it. Most public proof from training providers, ours included, sits in the first two levels. Your finance director cares about the last two.

Apply the same table to us. FuKazee has trained more than 3,000 people since 2022, by our own count. That figure covers corporate teams, postgraduate researchers and KCSE students together, so it tells a corporate buyer little about corporate work.

A six-point test for any recommendation letter

1. Does it name a person, a role and an organisation?
2. Does it say what was bought, for how many people and when?
3. Does it say who the audience was, and is that audience like yours?
4. How soon after the training was it written? A letter written within weeks can speak to delivery. It cannot speak to lasting change.
5. What is the relationship between the signer and the provider? A host, a friend or an investor may be sincere and still be close.
6. Can you confirm it with the signer? Our published letters hide signatures for privacy, so a careful buyer should confirm them directly, as with any letter that cannot be checked on its face.

Percentages deserve the hardest questions

If a provider quotes a productivity gain, ask what was measured, by whom, over what period, against what baseline and for how many people. Ask whether the checking and correcting time was counted. Ask to see the working. Put these questions to any such figure on our site as well.

What should you ask a provider's past clients?

Ask what staff were still doing three months later, and what went wrong. Try to speak to two people: someone who approved the spending and someone who sat in the room.

About the purchase

1. What exactly did you buy, for how many people, over how many days?
2. Who taught it? Was that the person who sold it to you?
3. Which version of the tools did staff use, and could they keep using them afterwards?

About fit

4. What did the provider ask before designing anything?
5. What changed in the plan after those conversations?
6. Did the exercises look like your work? Did any confidential material go into a tool during training?

About what lasted

7. Three months on, which tasks are people still doing with AI? How do you know?
8. What follow-up happened, and who started it, you or them?
9. Which materials do people still open?

About what went wrong

10. What disappointed you?
11. Did the provider ever tell you that training would not solve a problem?
12. Have you bought from them again? If not, why not?
13. What kind of organisation should not hire them?

Be wary if every reference is a coordinator who booked the room, if the referee repeats marketing language, or if nobody can name one thing that went wrong, and warier still if a provider connects you with nobody.

An email you can send to any provider

Thank you for your proposal. Before we decide, we would like to speak to two past clients: one person who approved the training and one person who attended it. Please suggest organisations whose staff and work are close to ours, and tell us what each one bought, how many people attended, when it ran and who taught it. If a client prefers not to take a call, a short written reply to our questions is fine. We would like to speak to them by [date].

A prompt for checking a provider's public evidence

Use the four Cs we teach: clarity, context, constraints and customisation. Paste public material only.

Review the public evidence below from an AI training provider, as a sceptical buyer. Context: I am an HR manager at a [sector] organisation in Kenya with [number] staff. We want staff to use AI on [two or three real tasks], and we care about what changes after three months. Constraints: use only the text I paste, add no outside facts, and write "not stated" wherever the evidence is silent. Customisation: give me a table with the columns Claim, Who says so, What was actually bought, What it does not prove and Question to ask. End with the five questions most worth asking this provider's past clients. Text: [paste web page text, published letters and public reviews]

AI can miss things or misread a letter, so read the source yourself before you rely on the table.

Which checks should HR run before signing?

Check the supplier documents, who will teach, how data is handled and whether staff can use the tools afterwards. Our procurement checklist and RFP guide go deeper.

Supplier documents. Our vendor information page lists what we supply on request, including the certificate of incorporation, KRA PIN certificate and bank confirmation letter.

Who teaches. Our About page says Rodney Maina teaches in every programme. Confirm it for your dates, and ask any provider who would step in if the named trainer could not.

What "your real work" means. Providers, including us, say they train on your work, and that can mean two very different things. At Maisha Girls Safe House, the organisation shared three of its document formats and we used them for structure only. Our vendor page says exercises run on synthetic material. Where an engagement involves your internal documents, ask for a confidentiality agreement first.

Tool access. Confirm before training that staff are allowed to use the tools on their work computers. We explain why this matters in why corporate AI training does not stick.

Where is FuKazee's evidence thin?

Our public evidence is strongest on delivery and weakest on lasting results.

• One published letter describes an employer training its own staff. Another concerns curriculum design, not a training room.
• All three letters date from August to October 2026. We have not published a follow-up showing what staff were doing six months later.
• We have not published an independent evaluation of our training.
• None of the three letters comes from a bank, a government body or a large listed company. If you are one of those, ask for a reference closer to your world.
• One lead trainer limits how many dates and rooms we can cover at once.

What does FuKazee not do?

FuKazee does not sell software or promise results. The full list comes from our public pages and from how we teach.

• We do not sell software. We help teams compare tools and do not promote any single one. We show the free version of each tool, and we cover Microsoft 365 Copilot alongside ChatGPT and Gemini.
• We do not need your confidential data to run a good session. No client's sensitive records should go into an AI tool during training.
• We do not promise results. Our site says that past learners' experiences are no promise of the same result, and that outcomes depend on each person's effort and circumstances.
• We do not claim anyone becomes an expert in a few days. We assume nobody in the room knows AI yet, and we teach clear instructions and how to check the output.
• We do not set out to produce machine-learning engineers or software developers. The aim is better everyday work for people in ordinary roles.
• We do not give a price before scoping. A scope call comes first, then a written quotation.

If you need a software vendor, a guaranteed return or specialist engineering training, look elsewhere or ask us early so that neither of us wastes time.

Questions people ask

Can we speak to FuKazee's past clients before we sign?

Ask us, and ask for someone whose work resembles yours. Our vendor page shows three recommendation letters and says references can be seen before engaging us. Whether a past client takes a call is that person's decision.

Did FuKazee teach AfricaHackon's learners?

No. FuKazee designed the curriculum, with AfricaHackon's feedback. We did not teach it, so we claim no learner results.

What is FuKazee's Google rating?

As of 8 November 2026, our reviews page shows 4.9 from 38 Google reviews. Check Google's listing for the current number.

Does a recommendation letter prove the training worked?

It proves less than it appears to. A specific letter shows who bought what and how they felt soon afterwards. Only a conversation with the client, or your own before-and-after measurement, can show what changed.

How long after training should we judge whether it worked?

We suggest three months. By then the excitement of the session has passed and you can see which habits have stayed. Record how long the chosen tasks take before training, so that you have something to compare.

Does FuKazee guarantee results?

No. Results depend on each person's effort, on manager support and on whether staff can use the tools at work. Agree how you will measure the result before training starts, with whichever provider you choose.

Sources

Last checked 8 November 2026.

FuKazee's public pages

• FuKazee reviews: client letters and Google reviews
• About FuKazee
• Vendor information for procurement teams
• AI training in Kenya for companies and teams
• Maisha Girls Safe House recommendation and training account
• HackHouse Africa recommendation and training account
• AfricaHackon recommendation and curriculum account
• FuKazee on who its corporate training is for (Clarity Corner)

Outside sources

• Google, Maps user-generated content policy: prohibited and restricted content
• Kirkpatrick Partners, The Kirkpatrick Model

Where FuKazee fits in

FuKazee trains teams to use AI on the work in front of them. We assume nobody in the room knows AI yet, we avoid jargon, and we teach the four Cs before moving on to CRAFT. Our corporate AI training page sets out the modules, formats and how a session is scoped.

If you are comparing providers, send us your questions from this piece and tell us what your teams do and how many people are involved. To start, email us at lead@fukazee.com