AnswersPublished14 sources
AI Agents Are Spreading Across Southeast Asia, but Most Teams Aren’t Ready for Autonomy
In Singapore, 83% of developers expect AI agents to handle most development work within three years, but only 37% say their codebases are ready for full autonomy. Across the surveyed markets, cost, integration complexity and governance are major barriers to expanding agent use.
Published byEdited with GPT-6 LunaImages generated with GPT Image 2
Research answer
Key takeaways
- In Singapore, 83% of developers expect AI agents to handle most development work within three years, but only 37% say their codebases are ready for full autonomy.
- Across the surveyed markets, cost, integration complexity and governance are major barriers to expanding agent use.
- The report also points to changing skill needs and greater career insecurity among junior developers.
AI Prompt
openai.comCreate a landscape editorial hero image for this Studio Global article: What does Agoda’s survey of over 800 actively employed software developers and engineering leaders across India, Indonesia, Malaysia, the Ph. Article summary: Agoda’s AI Developer Report 2026 reveals that AI-agent adoption is advancing faster than organisations’ readiness for fully autonomous development: agents are entering real workflows, but human oversight, cost controls, . Topic tags: general, general web. Style: premium digital editorial illustration, source-backed research mood, clean composition, high detail, modern web publication hero. Use reference image context only for broad subject, composition, and topical grounding; do not copy the exact image. Avoid: logos, brand marks, copyrighted characters, real person likenesses, fake screenshots, UI text, readable text, watermarks, charts with fake numbers, clic
Agoda’s AI Developer Report 2026 points to a gap between AI-agent adoption and readiness for autonomous software development. A survey of more than 800 developers and engineering leaders across India, Indonesia, Malaysia, the Philippines, Singapore, Thailand and Vietnam found agents entering real workflows, while human approval, cost controls and governance remain important.5
17
Singapore makes the contrast especially clear: developers are optimistic about what agents may do within three years, but far fewer say their codebases are ready for agents to work autonomously.7
Adoption is growing, but autonomy is not the norm
Across the surveyed markets, 53% of developers reported AI agents in production or broad use across their organisations. The report’s regional findings also show that adoption does not mean agents are operating without oversight: 79% of respondents said human approval is required for production deployment.17
In Singapore, 58% reported agents in selected production workflows, while 45% said agents were used broadly across teams—the highest share for broad use among the surveyed markets, compared with 31% regionally.7 These figures describe different kinds of use; neither indicates that agents have taken over development end to end.
Singapore’s expectations outpace codebase readiness
In Singapore, 83% of developers expect agents to handle most development work within three years. Yet just 37% described their codebases as mostly or fully ready for a fully autonomous agent.7 That 46-percentage-point difference separates a forecast about future work from an assessment of current readiness; it is not evidence that full autonomy is already achievable.
The practical challenge is preparing the surrounding software environment and processes—not only adopting agent tools. Codebases, validation practices and approval controls all matter when AI-generated changes are considered for production.7
17
Cost and integration complicate scale-up
The report identifies cost (28%), integration complexity (24%) and lack of governance (19%) as barriers to wider agent adoption across the surveyed markets.16
17 The cost concern extends beyond model calls: production agents also require work to design, integrate, review and validate them. Agoda CTO Idan Zalzberg has highlighted those broader requirements as part of building trustworthy systems.
17
That helps explain why lowering model costs alone may not remove the friction. Teams also have to account for the engineering and oversight needed to make agents useful within existing workflows.16
17
Usage controls and skills are part of the transition
AI use is increasingly managed through limits as well as deployment. In Singapore, 90% of respondents reported restrictions such as token quotas, usage caps or budget limits, compared with 81% regionally.16
17 These controls reflect the need to manage spending while teams expand their use of AI.
The report also points to growing importance for system architecture and design, AI literacy and agent orchestration. Career insecurity differs by seniority: 49% of junior developers expressed concern, compared with 17% of CTOs and VPs of Engineering.16
17 That suggests teams may need to address workforce concerns and skill development in ways that reflect different roles and levels of experience.
The leadership question is how to scale safely
For Singapore respondents, managing AI risk was the leading near-term challenge for engineering leadership, cited by 42%.7
17 Taken together, the findings suggest that scaling agent use involves more than increasing adoption: organisations must also develop governance, codebase readiness, validation and approval practices, alongside the skills to manage agent-assisted work.
These results describe surveyed developers’ reported practices and expectations. In particular, the expectation that agents will handle most development work within three years is a forecast from respondents, not proof that this level of autonomy will be reached.5
7
Studio Global AI
Continue your research
This page includes a source-backed answer you can continue inside Studio Global.
People also ask
What is the short answer to "AI Agents Are Spreading Across Southeast Asia, but Most Teams Aren’t Ready for Autonomy"?
In Singapore, 83% of developers expect AI agents to handle most development work within three years, but only 37% say their codebases are ready for full autonomy.
What are the key points to validate first?
In Singapore, 83% of developers expect AI agents to handle most development work within three years, but only 37% say their codebases are ready for full autonomy. Across the surveyed markets, cost, integration complexity and governance are major barriers to expanding agent use.
What should I do next in practice?
The report also points to changing skill needs and greater career insecurity among junior developers.