AI training for financial services in Malaysia works best when it is built around the work a bank actually does: reviewing documents, spotting anomalies in transactions, drafting customer correspondence and summarising policy. Generic prompt-writing workshops rarely transfer to a regulated environment. A programme that lands is one where every exercise uses a realistic banking or insurance task, and where governance, data handling and human review are taught alongside the tools themselves. Most in-house programmes of this kind can be structured to be HRD Corp claimable for employers who contribute to the levy.
This guide sets out what financial institutions in Malaysia should look for in an AI training programme, which use cases justify the investment, and how to think about governance before the first workshop runs.
Why financial services needs a different AI programme
Banks, insurers, takaful operators and capital markets firms sit under closer supervision than most industries. That changes what AI training has to cover. A marketing team can experiment with a generative AI tool and treat mistakes as cheap. A credit team cannot.
Three constraints shape the syllabus:
- Data sensitivity. Customer records, account data and internal risk assessments cannot be pasted into public AI tools. Staff need to understand where the boundary sits and which sanctioned tools are on the right side of it.
- Auditability. If AI assists a decision, someone will eventually ask how that decision was reached. Teams need habits that leave a trail.
- Accountability. Generative AI produces fluent output that can still be wrong. In a regulated setting, the person who signs off remains responsible regardless of which tool drafted the text.
Training that ignores these constraints produces enthusiasm followed by a quiet internal ban. Training that builds them in from the first hour produces adoption that survives the compliance review.
What Generative AI actually is, in this context
Generative AI refers to systems that produce new text, images, code or analysis from a prompt, rather than simply retrieving an existing record. In a financial institution the practical implication is that these systems are drafting and summarising assistants, not sources of truth. They are strong at restructuring information a human already has, and weak at supplying facts the human cannot verify. Getting a workforce to internalise that single distinction is often the highest-value hour in the entire programme.
Use cases worth training for
Document review and summarisation
Credit files, policy documents, legal agreements and regulatory circulars all arrive as long documents that someone must read and condense. This is the most reliable early win because the source material is in front of the reviewer, so accuracy can be checked immediately. Teams learn to extract obligations, compare two versions of a contract, and produce a summary in a fixed internal format.
Anomaly and pattern spotting
Fraud, AML monitoring and reconciliation teams work with exception queues. AI tools help draft investigation narratives, cluster similar cases and explain why a pattern looks unusual in plain language for a case file. The training emphasis here is on assisting the analyst rather than replacing the detection rules already in place.
Customer communication
Complaint responses, product explanations and branch correspondence take time to write well and consistently. Teams learn to draft to a house tone, adapt reading level, and produce Bahasa Melayu and English versions of the same message. The review step stays human.
Internal knowledge and onboarding
Operations and compliance teams sit on large internal procedure libraries. Training covers how to query internal documents effectively, and where retrieval-based approaches make more sense than asking a general model. This is often where the biggest quiet productivity gain sits.
Analyst and reporting support
Drafting board papers, converting a spreadsheet into commentary, and turning raw figures into a narrative are all tasks where a well-structured prompt saves hours. Finance and strategy teams benefit most.
Governance belongs in the training, not after it
Malaysia has an active national conversation about AI standards and responsible adoption, and most large financial institutions are drafting or refining internal AI usage policies. Rather than treat that as a separate compliance exercise, effective programmes fold it into the workshop itself. Participants leave having practised on realistic tasks and having applied the organisation’s own rules while doing so.
Practical elements to include:
- A clear list of approved and prohibited tools, taught by example rather than by memo
- A rule of thumb for what may and may not be entered as input
- A required human verification step for anything customer-facing or decision-relevant
- A simple record-keeping habit for AI-assisted work
- Escalation: who to ask when the rules do not obviously apply
For background on the wider Malaysian policy picture, see our guide to MY-AI standards and AI governance in Malaysia. Institutions should always confirm current supervisory expectations with their own compliance function, since requirements are updated over time.
How the programme is usually structured
A typical in-house engagement for a financial institution runs in three layers:
- Executive briefing (half day). Senior management and board-level stakeholders. Focus on what AI changes, what it does not, and what governance the institution needs in place.
- Core workforce workshop (one to two days). Hands-on, department-specific exercises using realistic tasks. Separate tracks for operations, credit, compliance, customer service and support functions where the work differs enough to justify it.
- Follow-through. A defined set of workflows each department commits to running with AI support, reviewed several weeks later. Without this step, most of the training decays.
Delivery is normally on-site at the institution’s own premises so that internal examples can be used safely. Details on formats and scheduling are on the corporate AI training page.
HRD Corp claimability
Employers registered with HRD Corp and contributing to the levy can generally use their levy balance for approved training programmes, including AI training delivered in-house. The programme, the provider and the trainer all matter to the application, so the claim route should be confirmed before the training dates are fixed rather than after. Our detailed walkthrough is here: How HRD Corp claimable AI training works.
About the trainer
Dr Muhamad Hariz Bin Muhamad Adnan holds a doctorate in Artificial Intelligence and is a Senior Lecturer at the Faculty of Computing and Meta-Technology, Universiti Pendidikan Sultan Idris (UPSI). He is an HRD Corp-certified AI trainer in Malaysia, working on AI-driven digital transformation in education and the workplace. Programmes are designed and delivered directly rather than passed to an associate pool, so the syllabus can be adapted to the institution’s own workflows.
Frequently asked questions
Do our staff need a technical background?
No. The core workshop is built for non-technical professionals in operations, credit, compliance, customer service and support functions. Technical teams can be given a separate deeper track if the institution wants one.
Can the training use our own documents and cases?
Yes, and it usually should. Working from sanitised internal examples is what makes the training transfer. The institution controls what material is shared and under what terms.
Is this HRD Corp claimable?
For levy-contributing employers, in-house AI training of this kind can normally be structured as a claimable programme. Confirm the specific claim route with HRD Corp before finalising dates.
How long before we see anything measurable?
Realistically, the first measurable change comes from a small number of repeated workflows rather than from broad tool adoption. Pick two or three tasks per department, measure the time they take before training, and re-measure four to six weeks after. That is a more honest signal than a satisfaction survey.
Can training be delivered in Bahasa Melayu?
Yes. Sessions can be delivered in English, Bahasa Melayu, or a mix, which is often the practical choice for branch and operations teams.
Next step
If you are scoping AI training for a bank, insurer, takaful operator or other financial services firm in Malaysia, the fastest way to get a realistic proposal is to share your team size, functions involved and preferred dates. Get in touch here and you will receive a syllabus outline and quotation you can take to your HRD Corp application.
