On this page
SACCO loan officers can use AI to draft member letters, write up loan appraisals from figures they have already checked, answer policy questions from the SACCO's own credit policy, and write simple money tips for members. AI does the writing. The officer checks it, and the credit committee and approved officers make every credit decision. Member records stay out of public AI tools.
This guide is for credit officers, loan officers and BOSA and FOSA staff in Kenyan SACCOs. It gives seven tasks with prompts you can copy, what to check, and what AI must never do. For a wider view of the sector, read our guide on AI for SACCOs and insurers in Kenya.
The rules that shape a loan officer's work
Two sets of rules matter most. Check both with your compliance officer, because they can change.
SACCO regulations. The Sacco Societies (Deposit-Taking Sacco Business) Regulations, 2010, as published on SASRA's website, require a written credit policy. Regulation 28 says it must cover, among other things, the requirements for a loan, the "appraisal of the borrower's ability to repay", approval limits for each officer and committee, and guarantor requirements. The same regulation requires sixty days' written notice to every member affected by a change in any term of their loan contract. Regulation 38 says that when a loan application is rejected, the SACCO must give the applicant the reasons in writing within fourteen days. These regulations apply to deposit-taking SACCOs. If yours is regulated differently, ask which rules apply.
Data protection. Member files hold personal data, and some of it is sensitive. Payslips, ID numbers, M-PESA statements and guarantor forms are all personal data. Some of what they hold, such as property details and family details, counts as sensitive personal data under the Data Protection Act, 2019.
As of November 2026, we found no AI-specific rule from SASRA. One draft is worth knowing. In March 2026, Kenya's financial regulators, including SASRA and the Central Bank of Kenya, published a draft Financial Consumer Protection Framework for public comment. Comments closed on 28 April 2026. As of 15 November 2026, only the draft appears on SASRA's and the CBK's websites, so it is not yet in force. It says computer programs, including AI tools, used on digital platforms should be risk-assessed, tested and documented. It also defines "plain language" as wording people with low financial confidence can understand, without legal or technical jargon.
Seven tasks where AI helps, with prompts
Each prompt follows the four Cs we teach at FuKazee: Clarity, Context, Constraints and Customisation. Replace the square brackets. Use invented or masked details: no names, ID numbers, phone numbers or account numbers.
1. Write a loan decline letter after the decision
I am a loan officer at a deposit-taking SACCO in Kenya. Our credit committee has declined a loan application. Write a respectful letter to the member that gives these reasons in plain English: [paste the committee's reasons]. Explain what the member can do next, for example reapply after [condition] or speak to the credit office. Constraints: under 200 words, no new reasons, no promises about future approval, and no legal threats. Leave [MEMBER NAME] and [DATE] as placeholders for me to fill in our system.
Check: the reasons match the committee minutes word for word in meaning. Send it within the fourteen days the regulations allow.
2. Explain a change in loan terms
Write an SMS of under 300 characters and a longer letter of under 200 words telling members about this change to our [loan product]: [paste the approved change]. Say when it takes effect and who to call with questions. Use plain English, then give a Kiswahili version of the SMS. Do not add anything that is not in the approved change.
Check: if the change affects existing loan contracts, the effective date gives affected members the full sixty days' written notice. A fluent Kiswahili speaker on staff reads the translation.
3. Write up a loan appraisal from checked figures
AI is useful for turning your checked numbers into a clear, consistent write-up. It must not do the checking, and it must not see the real member.
Using only the masked figures below, which I have already verified against the payslip and statements, write a loan appraisal note in our usual structure: applicant summary, purpose, income and existing deductions, repayment ability, deposits and guarantors, risks, and officer's comments. Show the repayment calculation step by step. Where a figure is missing, write [MISSING]. Do not recommend approval or decline. FIGURES: [paste masked figures, for example "Member A, net pay KES 62,000, existing loan deduction KES 9,500"]
Check: redo every calculation yourself. The recommendation is yours, under your credit policy and your approval limit.
4. Answer policy questions from your own credit policy
Below is our SACCO's approved credit policy. A member asks: "[question, for example: how many times my deposits can I borrow?]". Answer in under 80 words of simple English, using only the policy. Quote the section number. If the policy does not answer the question, say "The policy does not cover this. Please ask the credit manager." POLICY: [paste the relevant sections]
Check: the section number and the answer match. Save the good answers into a shared list your front office can use.
5. Explain guarantor duties before a member signs
Using only the guarantor clause below, write a one-page explainer for a member who has been asked to guarantee a colleague's loan. Include: what guaranteeing means, what could happen to their deposits if the borrower stops paying, how long the duty lasts, and three questions to ask before signing. Simple English, short sentences. CLAUSE: [paste]
Check: nothing in the explainer goes beyond the clause. The credit manager approves it before it is used.
6. Write respectful arrears reminders
Write three reminder messages for a member whose loan is [number] days in arrears. Message 1: friendly. Message 2: firmer, asks for a repayment plan. Message 3: explains the next step under our policy, which is [step]. Each under 60 words. Constraints: no threats, no shaming, no mention of the member's debt to anyone else, and no false statements about what will happen.
Check: regulation 38 bars a debt collector the SACCO appoints from harassing, oppressing or abusing anyone while collecting a debt. The draft consumer protection framework would also ban shaming, such as posting a person's debt on social media, and telling others about it except where the law allows. Keep messages private and respectful.
7. Write simple money tips for members
SASRA itself runs financial literacy outreach, including a March 2026 visit to a secondary school in Mombasa. Loan officers can support the same goal.
Write a four-week series of short WhatsApp tips for SACCO members on [topic, for example borrowing wisely, building deposits, or reading a loan statement]. One tip a week, under 70 words each, in simple English and Kiswahili. No product promotion, no promises of returns, and no figures unless I give them to you.
Check: every tip is accurate and matches your products. Your marketing or compliance lead approves before sending.
What AI must never do in a loan office
- Decide credit. In our view, an AI assistant should not approve, decline, score or rank loan applications. The credit policy sets who decides, and section 35 of the Data Protection Act gives people the right, with some exceptions, not to be subject to a decision based solely on automated processing that significantly affects them.
- See real member records in a public tool. Names, ID numbers, payslips, M-PESA statements, guarantor details and balances stay in your core system.
- Check documents for you. AI cannot confirm a payslip is genuine or that a guarantor really signed.
- Invent policy. If the answer is not in the pasted policy, AI should say so.
- Send anything unreviewed. Every letter and message is checked by a person before a member sees it.
How FuKazee trains SACCO teams
FuKazee runs AI training for SACCO credit, member service, finance and board teams, using invented or masked members throughout. Head office can train on-site by arrangement, with branches joining live online so counters stay open. We show free tools first, and Microsoft 365 and Copilot for SACCOs that already use them. See AI training for SACCOs in Kenya. Transport SACCOs may also like our guide on how a matatu SACCO can use AI.
Questions people ask
Can AI approve loans in a SACCO?
In our view, it should not. Loan decisions belong to the people and committees named in your credit policy. AI can prepare a clear write-up from checked figures. A person decides.
Can I paste a member's payslip into ChatGPT?
No. A payslip identifies the member and shows their income and deductions. Keep it in your core system. Type only the masked figures you need into an approved tool.
Does SASRA have rules on AI?
As of November 2026, we found no AI-specific SASRA rule. Existing SACCO regulations and the Data Protection Act still apply. Check SASRA's guidelines and circulars page for updates, and read our guide to the Data Protection Act and AI.
Can AI translate member letters into Kiswahili?
It can produce a good first draft. A fluent Kiswahili speaker on your staff should always read it, because a wrong word in a loan letter can mislead a member.
Sources
All sources last checked 15 November 2026.
- SASRA, Acts and Regulations: the Sacco Societies (Deposit-Taking Sacco Business) Regulations, 2010, regulations 28 (credit policy, notice of changes) and 38 (rejected applications, debt collection).
- SASRA, Guidelines and Circulars: the draft Financial Consumer Protection Framework for Kenya, March 2026 (definition of plain language; section 2.3.9 on digital platforms and AI; section 2.3.11 on debt collection). The same draft and public notice are on the CBK website.
- SASRA, "Empowering the Future: SASRA Takes Financial Literacy to Khadijah Secondary School, Mombasa" (March 2026).
- The Data Protection Act, No. 24 of 2019, ODPC copy: section 2 (sensitive personal data) and section 35 (automated decisions).
- FuKazee, AI training for SACCOs in Kenya.
Where FuKazee fits in
FuKazee is a Nairobi training company that has trained more than 3,000 people since 2022. We train SACCO teams to use AI on real member work, with masked records and clear rules on who decides. Tell us how many staff and branches you have and what you want AI to help with. 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.





