At an event built around agentic AI, Mixpanel’s Kurby Chua noticed something quieter than the conference theme. Relatively few product professionals were asking him about artificial intelligence.
“There’s not a lot of people mentioning to me about AI,” Chua said at the 2026 Indonesia Product Conference (IPC). His interpretation was not that Indonesian companies had ignored the technology. It was that their adoption was happening in a different layer: “A lot of the companies are just using AI, but as part of their work, not in terms of the product yet.”
That distinction is easy to miss when AI adoption is discussed as a single number. A product team can use assistants for research, code, analysis, or content while shipping no meaningful AI capability to customers. Internal use can make the factory faster without changing what leaves the factory.
Chua’s comments also sit beside a more optimistic judgment. He sees Indonesia developing a stronger product ecosystem, supported by active communities, broad app adoption, and an appetite for learning. The evidence suggests there is real momentum, but it does not support a simple claim that Indonesia is ahead of every comparable market.
Table of contents
Jump to each section:
- AI is visible at work, less so in products
- The product ecosystem has outgrown its starting point
- Regional data complicates a simple ranking
- What would move AI into the customer experience
- What this means for marketers
AI is visible at work, less so in products
Chua’s observation draws a useful boundary between AI as a work tool and AI as a product capability. The first can spread through individual subscriptions and team experiments. The second requires customer data, product design, reliability standards, governance, measurement, and a reason for the user to care.
Microsoft’s country findings from the Work Trend Index support the idea that workplace adoption is already visible in Indonesia. The equivalent Philippines findings point in the same direction, although at a lower reported level. 59% of Indonesian leaders and 44% of Philippine leaders said their companies were already using AI agents to automate workflows or business processes.
This is evidence of AI entering how companies operate. It is not evidence that the same companies have embedded AI into products customers use. The survey measures organizational workflows, which is precisely the layer Chua described.
The common assumption is that once product teams use AI, AI products naturally follow. The contrasting reality is that internal experimentation has a much lower threshold than customer-facing deployment. The strategic implication is that workplace adoption should be treated as a capability signal, not a product roadmap. AI inside the team is an input. AI inside the product is a promise to the customer.
The product ecosystem has outgrown its starting point
Chua’s broader assessment of Indonesia is rooted in what he encounters around the technology. “There’s a product community that is very strong and people are very eager to learn,” he said. Gatherings such as IPC matter because they are not only distribution channels for vendors. “It’s also for us to learn from the customer as well.”
That exchange matters in a market where capability is uneven. “There’s different levels of maturity for different product managers,” Chua said. A strong community does not erase those differences, but it can shorten the distance between them by circulating practices, exposing teams to other product models, and giving practitioners a language for problems they previously solved in isolation.
When asked about regional comparisons, Chua separated Singapore from the discussion because, in his words, “Singapore is way ahead.” His more revealing comparison was with his native Philippines: “Both countries started at the same pace in the past, but Indonesia now is way ahead.”
That is Chua’s assessment, based on the communities and product behavior he has observed, not a quantitative national ranking. He points to product communities he has not seen to the same extent in the Philippines and to Indonesians adopting a wider range of apps beyond digital wallets.
His explanation is cultural as much as technical. “Indonesians have a hunger to learn,” he said. “Once you’re open to learn, then you can advance, right, and you can improve your product or services that you’re offering.”
Product maturity often becomes visible before it becomes measurable. Communities create the social infrastructure for better products, but they do not by themselves prove that those products have crossed into advanced AI.
Regional data complicates a simple ranking
Independent data offers support for parts of Chua’s view, while pushing back on any broad conclusion that Indonesia simply leads the Philippines in AI adoption.
The e-Conomy SEA report from Google, Temasek, and Bain & Company shows strong consumer engagement with AI apps in Indonesia. 80% of Indonesian users covered by the report interacted with AI daily, indicating that AI services were already part of routine digital behavior.
That is a meaningful engagement signal. It does not tell us whether Indonesian product organizations built those services, whether AI sits inside their own core customer journeys, or whether activity is concentrated among a handful of global apps.
Microsoft’s Global AI Diffusion Report adds another complication because it uses a consistent measure of generative AI product use across economies. Generative AI diffusion reached 18.3% in the Philippines and 12.7% in Indonesia in the second half of 2025.
On that measure, the Philippines was ahead. The result does not invalidate Chua’s view of product communities or app ecosystems because it measures population-level use of generative AI products, not product management maturity. It does show why one observation cannot be stretched into a national AI hierarchy.
AWS research points to the same underlying problem in both countries: broad interest is not the same as deep integration. The Indonesia study, conducted in 2026, found most adopters were still exploring or experimenting, while a smaller group had fully integrated AI into operations and strategy. 56% of Indonesian AI adopters were exploring or experimenting, while 12% reported full integration into operations and business strategy.
The Philippines study, fielded in a different period, found a similarly narrow advanced layer. 78% of Philippine businesses remained focused on basic AI use cases, while 8% reached the most transformative stage where AI was core to product development, decisions, and business models.
Those AWS figures should not be read as a direct scorecard because the surveys were released a year apart and describe adoption stages somewhat differently. Their shared value is directional. In both markets, the distance between using AI and building around AI remains substantial.
What would move AI into the customer experience
Moving from workflow-level AI to product-level AI begins with a customer problem, not a model. Teams need to identify where prediction, generation, or autonomous action improves an existing journey enough to justify new uncertainty. An AI feature without a clear user benefit is merely an internal experiment moved onto the interface.
Product analytics then becomes more important, not less. Teams need to distinguish curiosity from repeat value, measure where users accept or override AI output, and understand whether automation removes friction or simply makes mistakes arrive faster. Chua’s presence at IPC is relevant here because learning from customers is part of the product work, not an activity that ends when a feature ships.
The difficult work also sits between functions. Product managers need engineers who can build reliable systems, legal and security teams that can define acceptable use, and marketers who can explain the capability without turning AI into the entire proposition. Customer-facing AI exposes the organization behind it.
The next maturity test is not how many employees have opened an AI assistant. It is whether a company can make AI useful, measurable, and accountable at the point where a customer encounters it.
What this means for marketers
For marketers, the distinction between AI in the workflow and AI in the product changes both positioning and evidence.
- Do not market internal adoption as customer innovation
A faster research process or content workflow may improve execution, but it does not create a customer-facing product benefit on its own. Claims should describe what changed for the user. - Build the message from product behavior
When AI enters a feature, marketers need evidence from adoption, repeat use, overrides, support issues, and customer outcomes. Novelty may earn attention, but observed value sustains the story. - Treat community as market intelligence
Chua’s description of Indonesia’s product community suggests that events such as IPC are listening environments as much as visibility opportunities. The questions practitioners do not ask can be as revealing as the ones they do. - Localize the maturity narrative
Regional demand for AI can coexist with uneven product readiness. Marketers should avoid importing a global AI storyline into Indonesia or the Philippines without checking how customers, product teams, and categories actually use the technology.
The deeper opportunity is not to declare a market ready for AI. It is to understand which layer is ready next.
Indonesia’s active product community may give companies a stronger learning loop, and its app economy shows clear demand. Chua’s surprise at IPC is a reminder that capability still has to travel from personal tools, through organizational systems, and into products customers can trust.
For marketing teams, that journey changes the job. The strongest AI story will not be the company saying it uses AI. It will be the customer recognizing that a product has become more useful, without needing the technology to be the headline.