AI will not fix a supply chain whose processes and data its own people have not first understood, simplified and agreed on. The firms that pull ahead will keep experienced people in charge of the judgement calls.
Big budgets, small returns
Southeast Asia’s digital economy is projected to reach US$368 billion this year (e-Conomy SEA 2026), and boardrooms from Singapore to Manila are backing AI with real conviction.
A February 2026 survey of 330 senior executives in six ASEAN markets, by McKinsey, Singapore’s Economic Development Board and Tech in Asia, found most organisations putting 11% to 40% of their technology budgets into AI. Yet 60% said it had added less than 5% to operating profit, and 18% saw no financial impact at all (EDB).
The pattern is global. IDC found that only four in every 33 AI pilots reach full-scale use (CIO), and Gartner expects more than 40% of agentic AI projects to be cancelled by the end of 2027 (Gartner).
This is not a software problem. It is about how the business is organised, how its processes run, how its information is kept, and who is accountable when a decision goes wrong.
Polished answers, shaky facts
AI returns a polished answer in seconds. Executives see the polish and assume the facts underneath are just as solid. Often they are not.
Freight platforms, warehouse systems and planners’ spreadsheets rarely agree, and AI does not settle those disagreements. It buries them inside a confident recommendation that only someone who knows the operation is likely to question.
“AI does not settle disagreements in your data. It buries them inside a confident recommendation.”
Planners feel it first. When they cannot trust the numbers, they pad safety stock and keep their own trackers on the side, tying up working capital and skilled time. Gartner predicts that through 2026, organisations will abandon 60% of AI projects not supported by AI-ready data (Gartner).
Clean data is a management job
Deciding what “on-time”, “delivered” or “delayed” actually means is a business decision, not an IT task. It needs operations, finance, customer service and compliance in the same room, with a leader senior enough to make the definitions stick.
ASEAN concluded negotiations on its Digital Economy Framework Agreement (DEFA) in May 2026, with signing expected at the 49th ASEAN Summit in November (Manila Bulletin). DEFA will make data easier to move across borders. It will not make that data accurate, and a company’s inconsistencies will surface in its partners’ customs declarations and audit trails.
For firms selling into Europe, the EU’s revised Product Liability Directive treats software, including AI, as a product for anything placed on the market after 9 December 2026 (EUR-Lex).
Fix the process before you automate it
Clean data is only half of it. AI runs whatever process it is given, at speed. A process full of workarounds, duplicate approvals and manual re-keying does not improve when AI is added. It gets worse, faster, and the mistakes travel further before anyone spots them.
“AI makes a good process faster. It makes a messy process worse.”
Before automating a workflow, map how it really runs, not how the procedure manual says it runs. Remove the steps that add no value, agree who owns each hand-off, and only then decide where AI can help.
What good looks like
- A simplified process, mapped as it really runs, with the workarounds removed before any AI is added.
- One agreed definition for each key measure, used by every team, system and AI tool.
- A named owner for every critical dataset, who signs it off before any AI tool relies on it.
- People in charge of exceptions. Any decision outside agreed limits goes to a named person, who owns the call.
Experienced people make it work
None of this comes out of a software box. Senior Subject Matter Consultants, practitioners who have run operations, trade compliance and supply chain teams, can bring operations, finance and IT to the same table and redesign the workflow before anyone writes code.
They also help leadership decide which decisions AI can make, which it can only recommend, and which a person must always own.
Before your next AI investment goes to the board, ask: Are our processes sound, is our data trusted, and are the right people still making the calls that matter?
If that question is on your board’s agenda, talk to us. Our specialists have run the operations they now help to fix, and we link our fees to measurable results.
Recent Comments