AI adoption is widespread, but developer confidence is still catching up, Agoda report finds

Artificial intelligence has become a routine part of software development across Southeast Asia and India, yet confidence in its reliability continues to lag behind usage, according to a new study by Agoda. The Agoda AI Developer Report 2025 shows that while developers are embracing AI tools for productivity gains, most still see them as assistants rather than dependable replacements for human judgement.

The report draws on survey responses from developers across the region and reflects an environment where adoption is no longer the main hurdle. Instead, the focus has shifted to trust, consistency, and the conditions required for AI to be used with greater confidence in production environments.

High usage, measured expectations

AI usage among developers in Southeast Asia and India is now close to universal. Nearly nine in ten respondents said they use AI tools on a weekly basis, with most reporting clear improvements in productivity and speed. These tools are commonly used to accelerate coding tasks, explore alternative solutions, and reduce repetitive work, making AI a regular feature of modern development workflows.

Despite this widespread use, expectations around AI capability remain cautious. Only 43 percent of developers believe AI can currently perform at the level of a mid-level engineer. This gap between adoption and confidence points to a pragmatic mindset, where AI is valued as an accelerator but not yet trusted as a replacement for experience or independent judgement.

Scepticism is evident across several major markets. Countries such as Thailand, India, the Philippines, and Singapore all show confidence levels that remain below or only marginally above the regional average. In each of these markets, a proportion of developers said they do not believe it is currently possible for AI to match mid-level engineer quality, indicating that the trust gap goes beyond short-term technical limitations.

Reliability concerns shape trust

The most significant barrier to deeper AI integration is inconsistent output. Seventy-nine percent of developers cited unreliable results as the primary concern, outweighing issues related to access, cost, or tooling. This suggests that availability is no longer the main challenge, and that predictability has become the defining factor in how AI tools are evaluated.

Concerns around reliability are especially strong in the Philippines and Thailand, where a large majority of developers highlighted output inconsistency as a major issue. Even in more mature markets such as Singapore and Malaysia, more than seven in ten developers remain cautious. This indicates that trust issues persist regardless of overall market maturity or exposure to AI technologies.

As a result, developers are careful about how AI-generated work is used. Rather than treating outputs as production-ready, many see them as drafts or starting points that require further validation. The report suggests that hesitation is driven less by resistance to AI and more by the need to manage quality and accountability in environments where errors can have serious consequences.

Human oversight remains central

To address these concerns, developers have adapted their workflows to retain control over outcomes. Two-thirds of respondents said they always review AI-generated code before merging it, and many routinely revise outputs until they meet internal quality standards. This emphasis on review highlights how AI has reinforced, rather than reduced, the importance of human oversight.

The findings show that AI adoption has increased the focus on verification and accountability. Developers continue to retain responsibility for final outcomes, with confidence in AI built gradually through testing, repetition, and experience. Trust, for now, is conditional rather than assumed.

Commenting on the findings, Agoda’s Chief Technology Officer Idan Zalzberg said the next phase of AI adoption would be defined not by who adopts AI first, but by who builds clear frameworks for consistent and productive use. He noted that while developers in Southeast Asia and India are advancing AI adoption with disciplined practices and peer oversight, the next step is to pair that momentum with stronger confidence in AI capabilities.

At Agoda, these insights reflect how engineering teams work with AI on a daily basis. Developers are encouraged to experiment with AI tools while maintaining clear ownership of outcomes, with review and verification remaining part of standard practice. This approach supports productivity gains while reinforcing confidence in the quality of AI-assisted outputs.

The Agoda AI Developer Report 2025 includes insights from developers across Southeast Asia and India, as well as perspectives from regional companies such as Carousell, MoMo, Omise, and SCB 10X. It presents a picture of a region that has moved beyond early adoption and is now focused on turning widespread use of AI into trusted, mature practice.