Agoda finds AI agents spreading across SEA software teams

Agoda has released an AI developer report on software teams in Southeast Asia and India. It found that 53% of surveyed developers now use AI agents in production or across broader organisational workflows.

Based on responses from more than 800 software developers and engineering leaders, the research points to a shift from AI tools that assist with individual tasks to systems that handle wider parts of the software development process. The survey covered India, Indonesia, Malaysia, the Philippines, Singapore, Thailand and Vietnam.

Developers reported larger productivity gains than a year earlier. Across the surveyed markets, 55% said they save at least seven hours a week through AI use, up from 18% in 2025.

Thailand recorded the highest share of developers reporting time savings of seven hours or more a week, at 70%. Vietnam followed at 63%, while the Philippines stood at 57%.

Use of AI-generated code also appears to be rising. Some 62% of respondents said AI-written code was often or almost always usable without major changes, although 86% said they still review or validate AI outputs always or most of the time.

Cost pressure

While AI agents are spreading, the findings suggest spending controls are shaping deployment decisions. Cost was the main barrier to broader adoption for 28% of respondents, ahead of integration complexity at 24% and lack of governance at 19%.

Four in five developers said they work under some form of AI usage limit, including token quotas, usage caps or budget restrictions. Spending on AI extends beyond model use to include design, integration, review and validation work.

Readiness for more autonomous systems remains uneven. Only 38% of those surveyed said their codebase is mostly or fully ready for a fully autonomous agent, even though 71% expect AI agents to handle most development work within three years.

Thailand led in reported AI agent use at 64%, followed by Vietnam at 59% and Singapore at 58%. Those figures refer to use in selected production workflows or wider organisational deployment.

Human checks

The research indicates that developers are more comfortable giving AI greater autonomy in lower-risk work such as documentation. As tasks move closer to live production systems, human sign-off becomes more common.

Some 79% of respondents said human approval is required for production deployment. That figure rose to 84% in both Vietnam and Thailand.

The results also suggest AI is changing the mix of skills expected from software teams. System architecture and design, AI literacy and agent orchestration were identified as areas growing in importance as developers work alongside more autonomous tools.

That shift is not affecting all groups in the same way. Nearly half of junior developers, or 49%, said they felt less secure about their future career prospects, compared with 16.7% of chief technology officer and vice president engineering respondents.

Idan Zalzberg, chief technology officer at Agoda, commented on the findings.

"Artificial intelligence is reshaping how developers across Southeast Asia and India build, learn, and collaborate. What began as a way to speed up individual tasks is becoming a broader change in how software is built. Across industries, developers are seeing stronger productivity gains and are beginning to use agents across more of the development lifecycle. But greater capability brings new responsibilities. The real cost is not simply the model call. It is the work required to make AI trustworthy in production. The organizations that will excel in this next phase will be the ones that manage cost carefully, build the right safeguards, and keep human judgment where it matters most," Zalzberg said.

The study was carried out with Macramé Consulting through an online survey conducted between June and August 2026. Respondents came from a range of experience levels, company sizes and sectors. The research also drew on a separate executive questionnaire and company case studies.

The findings underline a central tension in software teams' use of AI across the region: companies are widening deployment and developers are reporting time savings, but most still want people to check outputs before software reaches production.