Apiary Co-Founder and CEO Jessica Casey Jaya sees a clear change in the questions Indonesia’s product community is asking about artificial intelligence. At last year’s Indonesia Product Conference (IPC), she said the discussion was still largely an introduction to AI and the tools entering the market. This year, the emphasis has moved to what teams can actually build and operate with agents.
That shift shaped IPC 2026, held in Jakarta on September 24 around the theme of building products with Agentic AI. It also reveals a more consequential question beneath the event agenda. Once teams move from experimenting with AI to integrating it into products, who decides what those systems may do, how risk is controlled and who remains responsible when they fail?
Jaya believes that governance conversation has not yet received enough attention. Her observation does not establish a national adoption trend on its own, but it offers a useful view from an organization that researches its audience before setting the conference program.
Table of contents
Jump to each section:
- From AI orientation to product integration
- Governance becomes the unanswered question
- Apiary treats community demand as an editorial signal
- What product and marketing teams should take from the shift
From AI orientation to product integration
Apiary does not choose the IPC theme solely by following the loudest technology trend. Jaya said the organization conducts research before each event and tests its findings with experienced product leaders.
“Before every IPC, we conduct market research to understand what topics participants need and are looking for. We also have a board of advisers who are experts in their respective fields, particularly product, and we always consult with them.”
That process led the team to Agentic AI, with a specific interest in how agents can help product teams expand and connect AI capabilities to the products they already manage.
“That’s how we arrived at Agentic AI this year, and how people use AI agents to grow their products.”
The more revealing insight is how quickly the baseline changed. The 2025 conversation, in Jaya’s telling, focused on orientation. By 2026, the community wanted to discuss application and integration.
“Last year, Agentic AI wasn’t really being discussed yet. It wasn’t that hot. It was more like, ‘welcome to the world of AI.’ This year it’s more about the how-to, about growing and integrating AI with the product.”
One year can be enough for a technology conversation to move from vocabulary to operating choices.
The official IPC program reflects that transition. Its sessions focus on agents that can take action, the changing work of product managers, and the operational demands of keeping AI products running after launch. This is more specific than general enthusiasm for AI. It suggests that at least within the community Apiary convenes, access to tools is no longer the only constraint.
Governance becomes the unanswered question
When asked what remains under-discussed, Jaya’s answer was concise: “I think it’s more around AI governance.”
That gap matters because agentic systems change the nature of product risk. A tool that drafts an answer creates a review problem. A system that acts across workflows, data and customer touchpoints creates questions about authority, escalation and accountability. Product teams cannot resolve those questions through model selection alone.
The common assumption is that implementation maturity follows technical capability. The reality is that organizations can connect AI to a product before they have agreed on decision rights or acceptable failure. The strategic implication is that governance may become the limiting layer precisely when adoption begins to feel practical.
Jaya framed Indonesia’s position cautiously and as her own assessment, especially in sectors where errors carry serious consequences.
“Everyone is moving aggressively to implement AI right now, but in Indonesia we’re still not moving as aggressively as markets overseas. We’re still somewhat halfway there because, from the government side, there’s also concern about getting it wrong. Especially in banking, one mistake can be fatal.”
Her comparison should not be read as a measured national benchmark. It points instead to a tension product leaders will recognize: moving more slowly can look like an adoption disadvantage, yet the cost of an unchecked automated decision can make caution rational.
Governance is not the opposite of growth. For products that depend on trust, it is part of the product architecture.
Apiary treats community demand as an editorial signal
Apiary’s wider purpose helps explain why the year-on-year shift is worth watching. Jaya described the organization as a platform for professionals and business owners to discuss developments, exchange opportunities and develop new skills.
“Our main purpose is to facilitate professionals and business owners in Indonesia so they have a platform to discuss, exchange opportunities and upskill.”
The organization currently runs the Future of Work-focused HR conference and IPC. Jaya said it is considering additional conferences next year, although no subject or timing has been confirmed. The intent is to keep the agenda responsive rather than lock the organization into a fixed set of themes.
“If there’s a new trending topic, we’ll try to address it.”
This makes IPC a useful listening point, but not a proxy for every Indonesian product team. Its agenda is shaped by the needs reported by its participants and advisers. That boundary makes the signal more credible: it says something specific about what one active professional community wants to solve now.
Events often reveal technology maturity through the questions people are willing to spend time discussing. Introductory sessions indicate a need for shared language. Implementation sessions indicate a need for operating knowledge. Governance discussions, when they arrive, indicate that the consequences have become real enough to require collective rules.
What product and marketing teams should take from the shift
The move from AI experimentation to agent integration changes what responsible implementation looks like, especially when agents begin touching customers, campaigns or business decisions.
- Implementation needs a permission model
Before an agent is connected to a live workflow, teams should define what it can recommend, what it can execute and what still requires human approval. For marketers, that could mean separating an agent that drafts campaign ideas from one that changes CRM segments, publishes content or sends customer communications. - Risk depends on consequence, not novelty
An automated research summary and an agent changing an offer or customer message should not pass through the same controls. Governance should reflect how visible, reversible and consequential each action is. - Marketing workflows need clear ownership
Once agents begin working across content, CRM, customer data or campaign operations, teams need to know who owns the output. If something goes wrong, responsibility cannot sit ambiguously between marketing, product, data and the vendor. - Growth and control have to mature together
Agents can help teams move faster, but scale also increases the reach of a bad prompt, weak data input or unclear approval rule. The more customer-facing the workflow becomes, the more important those boundaries are.
The next phase of AI adoption will not be defined only by whether teams can make agents useful. It will be defined by whether organizations can make their use legible, bounded and accountable. For marketers, that means AI governance is becoming less of a technical discussion and more of an operating one. The question is no longer only what an agent can do, but what marketing is comfortable allowing it to do on the brand’s behalf.
For Indonesia’s product community, that may be the real distance between a “welcome to AI” moment and durable product practice. Learning the tool is an individual capability. Governing what it does is an organizational one.
That is where the conversation can widen beyond product teams. Marketers, customer experience leaders, legal teams and executives all shape the promises and decisions an AI-enabled product makes. Governance becomes meaningful when those functions agree on the boundaries before a system tests them in public.