Across Southeast Asia and India, a quiet transformation is underway: developers have made artificial intelligence a fixture of daily work, yet the institutions meant to guide that work have not kept pace. A 2025 survey by Agoda spanning more than 600 developers across seven markets reveals that near-universal adoption conceals a deeper unevenness — AI is embraced where risk is low and quietly set aside where consequences grow serious. The region now faces a question that follows every technological rush: not whether to adopt, but how to govern what has already been adopted.
Governance, Not Adoption, Will Define AI's Next Phase in Southeast Asia
Developers are comfortable with AI when the risk is low. They pull back when things get complicated.
Mark
So the headline is that 95% of developers use AI weekly. That sounds like the story is over—AI has won, everyone's using it. But you're saying that's not the real story?
Mimi
Right. The 95% number is real, but it's also misleading. What matters is what they're using it for. Ninety-four percent use it for code generation, but only 36% use it for deployment. That's a massive gap. Developers are comfortable with AI when the work is routine and the risk is low. They pull back when things get complicated.
Luke
How do we know that's about comfort versus capability? Could it just be that AI tools are genuinely better at code generation than they are at deployment?
Mimi
That's a fair question. The report doesn't separate those things out. But the fact that 60% of organizations have no formal AI policy suggests it's not just about tool capability. If there were clear governance, you'd expect more consistent adoption across the workflow.
Mark
So the governance gap is the real story. Only 25% of teams have official guidelines.
Mimi
Exactly. And what's filling that gap is peer review and developer discipline. Sixty-seven percent always review AI-generated code before merging. Seventy percent rework the outputs. That's good practice, but it's not scalable. It's culture, not process.
Luke
But wait—if developers are reviewing and reworking everything, and the code still functions correctly less than half the time without revision, doesn't that suggest the discipline is working? The code is getting fixed before it ships.
Mimi
It's working as a safety net. But it's also a sign that the tools aren't reliable enough to trust at scale. And the rework is invisible to the organization. You can't measure it, you can't learn from it, you can't improve it.
Mark
So Agoda's GenAI Proxy is the answer—it creates visibility and lets you measure what's actually happening.
Mimi
It's one answer. It lets you see adoption patterns, understand real usage, and refine policies based on actual practice. That's the infrastructure that turns adoption into governance.
Luke
But the report is from Agoda, and Agoda is the company building this solution. How much of this framing is about what the data actually shows versus what Agoda wants to sell?
Mimi
That's a real tension. The data about adoption rates and governance gaps is solid—it's from 600 developers across seven markets. But the conclusion that governance infrastructure is the answer is shaped by Agoda's own business interests.
Mark
So the reader should know that this is a company diagnosing a problem it's also positioned to solve.
Luke
Exactly. The governance gap is real. Whether Agoda's specific approach is the right solution—that's still an open question.
Il Polso
- 95% of developers across Southeast Asia and India use AI tools weekly, yet adoption collapses from 94% for code generation to just 36% for deployment — revealing a technology embraced at the surface but distrusted at the edge.
- 60% of organizations have no formal AI policy, leaving developers to improvise their own guardrails through peer review and manual reworking of outputs that function correctly less than half the time without significant revision.
- The absence of institutional structure has produced a fragile workaround culture — disciplined and peer-driven, but impossible to measure, scale, or systematically improve.
- Companies like Agoda are moving to close the gap with governance infrastructure such as a GenAI Proxy that authenticates usage, monitors costs, and turns invisible developer behavior into legible, refinable policy.
- Agoda's CTO has named governance — not adoption speed — as the next competitive differentiator, signaling that the region's AI maturity will be measured by the quality of its frameworks, not the size of its uptake.
Across Southeast Asia and India, a quiet transformation is underway: developers have made artificial intelligence a fixture of daily work, yet the institutions meant to guide that work have not kept pace. A 2025 survey by Agoda spanning more than 600 developers across seven markets reveals that near-universal adoption conceals a deeper unevenness — AI is embraced where risk is low and quietly set aside where consequences grow serious. The region now faces a question that follows every technological rush: not whether to adopt, but how to govern what has already been adopted.
Across Southeast Asia and India, artificial intelligence has moved from experiment to everyday necessity. A 2025 report from Agoda, surveying more than 600 developers across Indonesia, Malaysia, Singapore, Thailand, the Philippines, Vietnam, and India, found that 95% now use AI tools weekly — ChatGPT, Cursor, and GitHub Copilot woven into the rhythm of daily work. The numbers are striking. But they conceal a more complicated story.
Adoption is uneven in ways that matter. Nearly all developers use AI for code generation, but the numbers fall sharply as tasks grow more consequential — 57% for testing, 56% for planning, and just 36% for deployment. Developers are reaching for AI where it feels safest and pulling back where the stakes rise. The technology has spread fastest where errors are most forgiving.
What makes this pattern revealing is what it exposes about governance. Only one in four teams operates under official AI guidelines, and 60% of organizations have no formal policy at all. In the absence of institutional direction, developers have built their own guardrails: changing code review processes, always reviewing AI-generated output before merging, routinely reworking what the tools produce. It is peer accountability born of necessity. But most developers report that AI-generated code functions correctly less than half the time without significant revision — meaning the discipline is a workaround, not a solution. It cannot be measured or refined. It is culture, not process.
Leading companies are beginning to close this gap. Agoda is building a GenAI Proxy — a unified layer between developers and AI tools that authenticates usage, enforces limits, monitors costs, and creates visibility into how the tools are actually used. The goal is to turn informal practice into structured, scalable governance.
Agoda's Chief Technology Officer, Idan Zalzberg, framed the challenge directly: the next differentiator will not be who adopts AI first, but who builds a clear framework around it. The region has speed and adaptability. What it needs now is the institutional foundation to make that momentum reliable. The next phase of AI maturity in Southeast Asia and India will not be counted in adoption rates. It will be defined by how well organizations govern what they have already embraced.
Across Southeast Asia and India, artificial intelligence has moved from experimental tool to everyday necessity. A new report from Agoda, the digital travel platform, surveyed more than 600 developers across seven markets—Indonesia, Malaysia, Singapore, Thailand, the Philippines, Vietnam, and India—between August and September 2025, and found that 95% of them now use AI tools on a weekly basis. ChatGPT, Cursor, and GitHub Copilot have become woven into daily work. The adoption numbers are striking. But they mask a more complicated reality: the region's developers have embraced AI unevenly, and the institutions meant to guide that embrace have largely failed to keep pace.
The adoption cliff is visible in the data. While 94% of developers use AI for code generation—the most straightforward application—the numbers drop sharply as tasks grow more complex. Only 72% use it for documentation, 67% for code review, 57% for testing, 56% for planning, and just 36% for deployment. This is not a story of universal integration. It is a story of developers reaching for AI where it feels safest and pulling back where the stakes rise. The technology has been adopted fastest where the work is most routine and the consequences of error are most contained.
What makes this pattern significant is what it reveals about governance. Only one in four teams—25%—operate under official AI guidelines. Six in ten organizations, 60%, have no formal AI policy at all. Developers are navigating the integration of a powerful new tool largely without institutional direction. Instead, they have built their own guardrails. Fifty-seven percent report that they have changed their code review processes since adopting AI. Two-thirds always review AI-generated code before merging it into production. Seventy percent routinely rework the outputs the tools produce. This is peer accountability emerging from necessity, not from strategy.
Yet the quality problem persists. Most developers surveyed indicated that AI-generated code functions correctly less than half the time without significant revision. The tools are fast, but they are not reliable. Widespread adoption has not automatically produced reliable or consistent results. The discipline developers have imposed on themselves—the peer review, the reworking, the caution—is a workaround, not a solution. It keeps the code from breaking, but it does not scale. It cannot be measured. It cannot be refined. It is culture, not process.
Leading companies are beginning to close this gap through structured governance. Agoda, which commissioned the report, is building what it calls a GenAI Proxy—a unified layer that sits between developers and the AI tools they use. The proxy authenticates usage, enforces limits, monitors costs, and ensures compliance across multiple large language model providers. This infrastructure does more than control access. It creates visibility. Agoda can now measure adoption patterns, understand how developers actually use the tools, and refine policies based on real behavior rather than assumptions. The company positions this as the foundation for scaling AI responsibly as the technology continues to evolve.
Idan Zalzberg, Agoda's Chief Technology Officer, framed the challenge plainly: the next differentiator will not be who adopts AI first, but who builds a clear framework around it. Southeast Asia and India have speed and adaptability as strengths. Developers in the region are advancing AI adoption with peer oversight and disciplined practices. But the next step, Zalzberg argued, is to pair that momentum with stronger foundations in trust, quality, and governance. Companies that do so will set the standard for reliable and responsible AI across the region.
The report drew on input from developers across the region and case studies from leading regional companies including Carousell, MoMo, Omise, and SCB 10x. The picture that emerges is of a region in transition. AI adoption is nearly universal. But the institutions, policies, and frameworks that turn adoption into reliable practice are still being built. The next phase of AI maturity in Southeast Asia and India will not be defined by how many developers use these tools. It will be defined by how well the organizations they work for govern that use.
Citazioni salienti
The next differentiator will not be who adopts AI first, but who builds a clear framework around it for consistent and productive usage.— Idan Zalzberg, Chief Technology Officer at Agoda
Vuoi la storia completa?
Leggi l'originale su Lokmat Times ↗
Technology