AI & Digital
Where should Indonesian companies start their AI journey?
Most AI programmes stall because they start from a tool rather than a decision that costs money. A readiness-first sequence protects the budget and the credibility of the team asking for it.
The question executives ask is usually "which AI platform should we buy?" The more useful question is "which recurring decision in this company is expensive, slow, or inconsistent — and would better information change the outcome?" AI investment becomes defensible the moment it is attached to a decision with a cost attached to it.
Start where the money already leaks
In asset-intensive and procurement-heavy Indonesian organisations, the highest-value early candidates are rarely customer-facing. They tend to sit in contract obligation tracking, vendor performance comparison, spend classification, technical document review, maintenance history analysis and demand aggregation across sites. These are areas where the data already exists as a by-product of operations, and where errors are quietly absorbed rather than reported.
Test readiness before ambition
A use case can be commercially attractive and still fail, because readiness has several independent dimensions. Data has to be available, accessible and reliable enough for the decision. The underlying process has to be stable enough to automate or augment. Users have to be able to change how they work. Governance has to cover privacy, security and accountability before a model touches production data.
- Is the data available, accessible and usable?
- Is the process stable enough to automate or augment?
- Can existing architecture support the use case?
- Are users prepared and capable of adopting the change?
- Are privacy, security, accountability and AI controls sufficient?
Sequence the first twelve months
A workable first year usually looks the same across sectors: a short discovery to fix the problem statement, a structured readiness assessment across data, process, technology, people and governance, a prioritisation of use cases by value, feasibility and risk, one business case worth defending, then a narrow pilot with a measurement plan agreed in advance.
The pilot's purpose is not to prove that AI works. It is to prove that this organisation can capture value from it repeatedly. That is why adoption and capability development belong in the plan from the start rather than being added after go-live.
What to avoid
Three patterns account for most wasted spend: buying a platform before defining a decision, running many disconnected proofs of concept that never accumulate into capability, and treating governance as a legal formality at the end instead of a design input at the beginning.