On this page
In September 2026, FuKazee trained 13 staff at Maisha Girls Safe House to use AI for communication and reporting. On 1 October 2026, the organisation put its recommendation in writing. This article explains what the training covered, why we deliberately took it slowly, and shares the letter in full.
Who is Maisha Girls Safe House?
Maisha Girls Safe House (MGSH) is a registered NGO in Nairobi. It provides advocacy, shelter and care for girls who are survivors of sexual violence.
Alongside the daily care of the girls, a small team carries a heavy writing load: reports for partners and donors, newsletters to supporters, letters and everyday communication. Every one of those documents has a reader who decides something based on it.
What did MGSH ask FuKazee for?
MGSH wanted its team to use AI tools for reports, newsletters and everyday communication, and to do so safely.
The team was mixed. Leadership, finance, monitoring and evaluation, communications, psychosocial support, programme staff and interns all trained in the same room. That mattered: one generic course would not have fitted everyone, so the second day was built around each person's own work.
Why did we take this training slowly?
The information an organisation like MGSH holds is among the most sensitive any organisation can hold: survivor stories, legal cases, medical details and the locations of girls in care. So one condition shaped everything:
No survivor information, case detail or identifying material goes into an AI tool. Ever.
In practice, that meant:
- A discovery session first. Before any slides were written, we met the team to understand their work, their documents and their worries.
- Safety before tools. The rules on what never goes into an AI system were taught on the first morning, before anyone opened ChatGPT, and repeated on the second day.
- Fictional material only. Every exercise used invented or sanitised examples. MGSH shared three of its own document formats, and we used them for their structure only, never their content.
- Written rules to keep. We left a one-page list of AI dos and don'ts for management to review alongside MGSH's own data protection and safeguarding policies.
We also found something every NGO in this space should know. Several AI tools refuse prompts that contain phrases such as "sexual violence", even in clearly fictional exercises. We showed the team how to word requests so the work gets done without ever adding real case detail.
What did the 13 staff learn?
The training ran for a full day and a half-day session on 18 and 19 September 2026 in Nairobi, led by FuKazee founder Rodney Maina.
Day 1 built a shared foundation:
- What AI is, and where it fails (including why it can sound confident and still be wrong).
- How to write clear instructions: clarity, context, constraints and customisation first, then the CRAFT method.
- How ChatGPT, Claude, Gemini and Perplexity differ, using the free versions.
- How to protect the people behind MGSH's records.
Day 2 moved from learning to setting up:
- Each person mapped the writing they do every week or month.
- The team began drafting standard operating procedures (SOPs) for that work.
- Staff set up their own AI projects for their roles, covering areas such as monitoring and evaluation, handling incoming calls, supplier invoices, finance and project management.
What did MGSH take home?
As the letter below notes, the team left with resources to return to after the sessions:
- A training report for management.
- An AI Playbook with prompt templates for each role.
- A participant guide with practice exercises.
- Certificates for all participants.
- Fourteen days of follow-up support for questions.
What did Maisha Girls Safe House say?
"The training was practical and tailored to our organisation. Our team learned to use AI tools for reports, newsletters and everyday communication, and to do so safely, with careful attention to protecting confidential information, which is essential in a setting like ours."
Abraham Liona, Finance and Administration Manager, Maisha Girls Safe House
The recommendation letter in full
MGSH agreed to this letter being published. The signature has been hidden for privacy.
The text of the letter reads:
RE: Recommendation for FuKazee Limited
Maisha Girls Safe House is a registered NGO in Nairobi, Kenya (NGO: OP.218/051/20-129/11906) that provides advocacy, shelter and care for girls who are survivors of sexual violence. Alongside the daily care of the girls, our team writes reports for partners and donors and communicates with the sensitivity our work requires.
To strengthen this side of our work, we engaged FuKazee Limited to train 13 members of our staff on the practical use of AI for communication and reporting. The training took place on 18 and 19 September 2026 in Nairobi, and was preceded by a discovery session in which FuKazee took time to understand our work and our needs.
The training was practical and tailored to our organisation. Our team learned to use AI tools for reports, newsletters and everyday communication, and to do so safely, with careful attention to protecting confidential information, which is essential in a setting like ours. FuKazee also provided a training report, an AI Playbook with prompt templates for each role, and certificates for all participants, giving our staff clear resources to return to after the sessions.
We found FuKazee professional, well prepared and responsive throughout. We recommend them with confidence to any organisation seeking practical, relevant AI training for its staff.
Mr. Abraham Liona, Finance and Administration Manager, Maisha Girls Safe House
Planning AI training for your NGO?
If your organisation handles sensitive information, the way you train matters as much as the tools you choose. We wrote a practical guide from this work: how to train NGO staff on AI, safely.
To see how we run AI training for NGOs in Kenya, including donor reporting, proposals and M&E, visit the NGO page. You can also read every recommendation letter and Google review in one place.
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.
