On this page
Bank relationship managers can use AI to prepare for client meetings from public information, draft follow-up emails, structure proposals, tidy call notes and explain products in plain language. The rule that keeps it safe: customer data stays in the systems your bank has approved. Public AI tools get only public information or text with every customer detail removed.
This guide is for relationship managers in corporate, SME, business and personal banking in Kenya. It gives seven tasks with prompts you can copy, what to check, and what AI should never do. For the bank-wide view, read AI in Kenyan banking: what staff training needs to cover.
What the Central Bank of Kenya has said
AI is already in the sector. The CBK's Survey on Artificial Intelligence in the Banking Sector, published in 2025, collected data as of 31 December 2024 from 37 commercial banks, 14 microfinance banks, one mortgage finance institution, three credit reference bureaus and 70 digital credit providers. Half of the respondents said they had adopted AI. Of those, 68 percent said they were using generative AI tools. Leading risks they reported included data quality and governance, cybersecurity, and a shortage of AI-skilled staff. Among commercial banks, 92 percent said the CBK should issue guidance on AI.
Draft rules are on the way. In September 2026, the CBK published draft revised Risk Management Guidelines for public comment. Comments closed on 7 November 2026, and as of 15 November 2026 they are still a draft, not final. In the part on reputational risk, they include a short item on artificial intelligence systems. It says institutions should make sure AI systems operate "in a transparent, secure, fair, and accountable manner, with appropriate human oversight", with model validation, bias testing and continuous monitoring.
Plain language for customers. In April 2026, the CBK invited public comment on a draft Financial Consumer Protection Framework prepared with other financial regulators. Comments closed on 28 April 2026, and as of 15 November 2026 it is also still a draft. It would require institutions to explain the benefits, features, risks, terms and costs of a product in plain language, including before a customer signs a credit contract.
Data protection applies now. The Data Protection Act, 2019 is in force. The Act defines personal data as any information about a person who can be identified. So customer names, account details, financial statements and call notes about a named person are personal data. Your bank's own policy sets which AI tools you may use for which kinds of information.
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 public information or fully masked text only, unless your bank has approved a tool for customer data.
1. Prepare for a meeting from public information
I am a relationship manager at a bank in Kenya. I am meeting the finance director of [company name], a [sector] business, next week. Using only public information I paste below (their website, annual report extracts and recent news), give me: a five-line summary of the business, three likely money needs for a business like this, five open questions to ask, and anything in the news I should be aware of. Mark anything you are unsure about. PUBLIC INFORMATION: [paste]
Check: confirm every fact against the source. Public news can be wrong or out of date. Never paste the client's account history into a public tool.
2. Brief yourself on a sector
Explain the [sector, for example horticulture exports or private schools] business in Kenya to a relationship manager in under 400 words: how money flows through the year, common cash pressures, and the questions a banker should ask. Do not give statistics unless you name the source. Flag anything I should verify.
Check: treat it as a starting point. Ask your bank's research or credit team for the figures you will rely on.
3. Turn rough call notes into a clean record
Turn these rough notes into a meeting record with these headings: purpose, what the client said, needs identified, products discussed, agreed next steps with owners and dates, and open questions. Do not add anything that is not in the notes. Write [UNCLEAR] where the notes are unclear. NOTES: [paste notes with the client's name, company name, account numbers and figures removed]
Check: the record matches what was said. The final copy goes into your bank's customer relationship system, not a personal AI chat history.
4. Write the follow-up email
Write a follow-up email to a client after a meeting. Thank them, confirm the three next steps below with dates, list the documents we need from them, and say who will contact them next. Formal but warm. Under 180 words. No promises on pricing, approval or timelines beyond those listed. NEXT STEPS: [paste]
Check: nothing in the email commits the bank to a rate, limit or approval. Credit and pricing decisions come from the people who own them.
5. Structure a proposal
I am preparing a proposal for a [type of business] client who needs [need, for example working capital for stock and a way to collect payments]. Using only the approved product descriptions below, suggest a proposal structure: the client's situation in their words, the solution, how each product helps, what the client needs to provide, and next steps. Do not describe any feature, fee or rate that is not in the product descriptions. PRODUCT DESCRIPTIONS: [paste approved material]
Check: product terms match your latest approved sheets. Compliance or product teams sign off client-facing documents as your bank requires.
6. Explain a product in plain language
Explain [product, for example an invoice discounting facility] to a small business owner who has never used one. Under 150 words. Cover what it is, how it works, the main risks and what it costs in general terms, using only the approved description below. Simple words, no jargon. Then give a Kiswahili version. DESCRIPTION: [paste]
Check: the explanation covers risks as clearly as benefits. A fluent speaker checks the Kiswahili.
7. Prepare for a credit conversation
A client wants to discuss a [facility type]. List the documents and information a bank in Kenya usually asks for, and ten questions I should ask to understand the request properly. Do not assess the client or suggest whether it should be approved.
Check: compare the list with your bank's credit checklist. Your credit team and approval process decide. AI does not write the credit judgement.
If your bank has rolled out Microsoft 365 Copilot or another approved assistant, it may be allowed to work with internal documents. Ask your IT and compliance teams exactly what is permitted. Our Microsoft Copilot training covers how it works with your files.
What AI should not do for a relationship manager
- See customer data in a public tool. No names, account numbers, statements, ID numbers or credit files in free or personal AI accounts.
- Make or suggest credit decisions. Approval belongs to your credit process. Section 35 of the Data Protection Act also gives people the right, with some exceptions, not to be subject to a decision based solely on automated processing that significantly affects them.
- Quote rates, fees or limits. These change and must come from approved sources.
- Write unchecked client messages. Every email and proposal is reviewed by you before it leaves.
- Replace the relationship. AI can help you prepare. Trust comes from you listening and following through.
How FuKazee trains banking teams
FuKazee runs AI training for banks and financial institutions in Kenya, practised on banking work inside each institution's compliance rules. Exercises use invented borrowers and fictional complaints, so no real customer data is needed. Staff learn to sort information into what may go into a public tool, what may go only into an approved tool, and what must never leave core systems. See AI training for banks in Kenya, and for client-facing skills more widely, AI training for sales teams.
Questions people ask
Can I use ChatGPT for my client meeting prep?
You can use a public tool with public information, such as a company's website or published annual report. Do not paste anything from the bank's systems. Check your bank's AI policy first, because it may not allow public tools at all.
Has the CBK issued AI rules for banks?
As of 15 November 2026, the CBK has published draft Risk Management Guidelines that include a short item on AI systems. The comment period closed on 7 November 2026, but they are not final. Check the CBK website for the latest status.
Can AI write a credit memo?
AI can help structure a memo from figures an analyst has checked, inside a tool the bank approves. The judgement, the recommendation and the approval stay with people. Your bank sets its own rules on this, so follow them.
Is it safe to record client calls and have AI summarise them?
Only if your bank allows it, the client has been told, and the recording and summary stay in approved systems. Recording a call creates personal data, so the Data Protection Act applies.
Which skills should a relationship manager learn first?
Start with writing clear prompts, masking customer details, and checking every output against a source. Those three habits make every later tool safer and more useful.
Sources
All sources last checked 15 November 2026.
- Central Bank of Kenya, Survey on Artificial Intelligence in the Banking Sector: survey method, adoption, generative AI use, reported risks and the proposal for CBK guidance.
- Central Bank of Kenya, Draft CBK Prudential Guidelines, Risk Management Guidelines, Guidance Notes and D-SIBs Framework (September 2026): draft Risk Management Guidelines (item 11, "Artificial Intelligence systems", in the reputational risk part) and the public notice with the 7 November 2026 comment deadline.
- Central Bank of Kenya, Public Notice on the Draft Financial Consumer Protection Framework for Kenya (April 2026): draft framework, section 2.4.2 on explanations for consumers, and the 28 April 2026 comment deadline.
- The Data Protection Act, No. 24 of 2019, ODPC copy: section 2 (personal data) and section 35 (automated decisions).
- FuKazee, AI training for banks in Kenya.
Where FuKazee fits in
FuKazee is a Nairobi training company that has trained more than 3,000 people since 2022. We train banking teams to use AI on real client work, with invented data and clear rules on what never leaves the bank's systems. Tell us which teams you want to train 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.





