Sit in any boardroom or ministry briefing in Malaysia today and AI will come up within the first ten minutes. Leaders have seen the demos, staff use chat tools at their desks, and every strategy paper has a section on it.
Interest is high. The harder question is what happens after the meeting.
The awareness is already here
Look at the three places where national momentum shows up, and all three are moving.
- Policy. The National AI Action Plan 2026 to 2030 sets out 28 initiatives and a target of a top ten global ranking in AI readiness by 2030.
- Infrastructure. MIDA has recorded RM385.7 billion in data centre related investment between 2021 and the first half of 2026.
- Business. A study by Strand Partners for AWS found that 38% of Malaysian businesses now consistently use at least one AI tool, up from 27% a year earlier.
That is a country that knows AI matters.
Where it stalls
The same study shows where things slow down. Of the businesses using AI, 67% are still focused on basic applications such as public chatbots and ready-made tools. Only 19% have a formal strategy for scaling AI across more than one function. Just 30% have clearly defined who is accountable for their AI initiatives, and 27% monitor or audit them regularly.
Malaysia is not alone in this. MIT's NANDA initiative reported in 2025 that around 95% of the enterprise generative AI pilots it studied had no measurable effect on profit and loss. The main reason it gave was poor integration with the way work is actually done.
In the conversations we have with public sector bodies and financial institutions, the same four obstacles come up again and again.
- The pilot has no owner after the demo. A proof of concept impresses a committee, then waits for a budget line that nobody was asked to create.
- AI sits outside the system where work happens. Staff copy text from the case system into a chat window and paste the answer back. It saves a few minutes, and it leaves no record.
- The data is not ready. Applications, approvals and supporting documents live in spreadsheets, shared folders and scanned PDFs, so there is nothing clean for AI to work on.
- Governance arrives last. Who is accountable for a decision the AI influenced? How is it audited? For government and regulated organisations, these questions stop a go-live, so they need answers on day one.
What implementation looks like
Getting past the pilot stage is mostly a matter of discipline. The approach we use has four steps.
- Start with one workflow that has a cost you can measure. Hours spent screening applications, days lost waiting for reviews, errors found in audit. Pick one.
- Build AI into the existing system. The AI reads from and writes to the same records staff already use, so its work is part of the process and part of the audit trail.
- Keep a person accountable. AI prepares, checks and recommends. A named officer approves, and every step is logged.
- Measure, then widen. Compare the workflow before and after. When the numbers hold, apply the same pattern to the next workflow.
Our own platform, mvpAI, follows this pattern. It is in production for a licensed Malaysian financial services provider, built into their existing system. Each application is screened as it arrives, scored the same way whoever reviews it, and summarised so reviewers start from the key points. The reviewers still make the decision. They simply spend their time on judgement, with the routine checks already done.
The opportunity
The AWS study holds one more finding worth noting. 57% of Malaysian businesses get their AI mainly through external providers, and 69% say locally based software providers and AI developers are important to their adoption. 72% say they would be more likely to increase their own use of AI if the public sector adopted more of it.
Put together, that describes a large opportunity for Malaysian organisations that move from experiments to working systems, and for local teams who understand Malaysian rules, procurement and data protection well enough to build them properly.
The organisations that benefit most from AI will be the ones that pick one process and make it work, end to end, inside the systems they already run.
If there is a process in your organisation that AI should be helping with, tell us what it is. We will tell you honestly whether AI is the right fit, and what it would take to put it into production.


