The Klang Valley CFO Survey: AI in Finance Operations

What 60+ Klang Valley CFOs told us about where AI is replacing manual work in their finance teams — and the specific places it absolutely is not, despite the hype.

The survey

Between January and March 2026, our advisory team conducted structured interviews with 63 CFOs and Heads of Finance at Klang Valley organisations — mostly mid-cap (200-2,000 staff), spanning manufacturing, BFSI, professional services, and consumer goods. The brief was simple: where, specifically, is AI changing how your finance function operates?

The findings were less about AI capability and more about which capabilities matter for finance specifically. The summary: there is a narrow set of workflows where AI is genuinely altering finance ops in Malaysian mid-caps, a broader set where it is barely scratching the surface, and a small set where there is more vendor noise than actual deployment.

Where AI is materially changing finance work

• Accounts payable invoice processing. 71% of respondents had AI-assisted invoice capture in production. Median time-per-invoice down 55%. This is the most boring and most universal use case.
• Bank reconciliation. 48% had reconciliation tooling that uses ML for matching unstructured statement items. Median reconciliation cycle time down 40%.
• Accounts receivable follow-up. 35% had agentic dunning workflows in production. Median DSO improvement: 6-9 days depending on customer mix.
• Variance analysis on management reporting. 27% used LLM-assisted commentary drafting on monthly management accounts. Most still review heavily, but draft time down ~60%.

Where AI has not landed yet (despite the hype)

Two areas were striking for how little AI penetration exists, despite consistent vendor messaging.

1. FP&A scenario modelling. Only 12% had any AI involvement in scenario modelling. CFOs consistently said the same thing: "I do not trust the inputs enough to trust the outputs." Until the data layer beneath the scenarios is reliable, AI does not solve the problem.
2. Treasury and cash forecasting. 8% had AI in production for short-term cash forecasting. The bottleneck is not the model — it is access to consolidated visibility across multiple banking relationships, which is a data-engineering problem masquerading as an AI problem.

What the LHDN e-invoicing rollout changes

The most-cited near-term inflection point: the phased LHDN e-invoicing mandate. Once invoices are submitted in structured format to LHDN, that same structured data is available to the finance function in a way it has not been before. Several respondents said they were holding back on AI investment specifically until e-invoicing creates a clean data foundation to build on.

This is a sensible posture. AI workflows built on top of OCR'd PDF invoices will be obsoleted by workflows built on top of structured e-invoice feeds. If you can wait six months, you save a re-architecture.

The talent pattern

A consistent observation across respondents: AI in finance is not reducing headcount. It is reshaping role mix. Senior accountants are now spending more time on review, exception handling, and interpretation. Junior roles are increasingly hybrid — part traditional finance work, part operating-the-Workers.

I have not let anyone go. But my next three hires look different from my last three. I need people who are comfortable working with AI outputs, not just comfortable producing them.
— CFO, Klang Valley industrial-equipment distributor

What we recommend for Malaysian CFOs reading this

1. If you do not have AP invoice processing automated yet, that is your highest-ROI starting point. It is mature, well-understood, low-risk.
2. If you are considering FP&A or treasury AI investment, audit your underlying data layer first. AI on bad data produces bad answers faster.
3. Plan your hiring mix for the next 18 months around the AI-augmented operator profile, not the pure-accountant profile. The skill gap closing fastest is not Excel — it is judgement about AI outputs.
4. If the e-invoicing wave hits your sector in the next 12 months, sequence your AI investment around it. Build on structured data, not on patches over PDF chaos.

Request the full Klang Valley CFO survey deck →