AI speeds coding but slows Australian software teams

GitLab has reported that Australian software teams are adopting artificial intelligence at speed while becoming less efficient overall due to fragmented tools, compliance pressure and skills gaps.

The company released new research on DevSecOps practices from a global survey of 3,266 professionals, including more than 250 based in Australia. Respondents worked in IT operations, IT security or software development.

The study describes an "AI paradox" in software development. Coding tasks are faster with AI, but wider delivery processes are slowing down.

Lost time

The data shows that DevSecOps professionals in Australia lose an average of seven hours per week because of inefficient processes. The report links this mainly to the number and variety of tools in use across teams.

Thirty-five per cent of respondents cited too many tools as a key cause of inefficiency. Thirty-one per cent pointed to a lack of knowledge sharing. Another 31% blamed different teams using different tools.

Sixty-seven per cent of respondents said they use more than five development tools. The survey found that 63% use more than five AI tools. That level of AI tool use is 14% higher than the global average in the study.

Views on future architectures are also shifting. Eighty-two per cent of respondents said agentic AI will only succeed within a platform engineering approach.

Roles in flux

The report indicates that AI is reshaping roles across DevSecOps teams. It also shows that demand for skills is rising as adoption spreads.

Eighty-two per cent of those surveyed agreed that as AI makes coding easier, the industry will employ more engineers, not fewer. Eighty-nine per cent said adopting AI is essential for future-proofing a software career.

Most respondents want more investment in training. Eighty-seven per cent said they want greater investment in upskilling. Eighty-four per cent expect their role to change significantly within five years.

The findings suggest that AI is becoming a standard part of software work in Australia. They also show that many professionals feel pressure to adapt their skills.

AI everywhere, but watched

AI use is now almost universal in the software lifecycle, according to the research. Ninety-nine per cent of respondents said they are using or planning to use AI across development, security and operations tasks.

However, teams are cautious about leaving work to AI without checks. On average, respondents trust AI to handle only 39% of daily tasks without human review.

Seventy-eight per cent said they have encountered problems from what they described as "vibe coding". This refers to code produced by AI that appears plausible but contains errors or omissions.

Most respondents see limits in automation. Eighty-eight per cent said human creativity and innovation cannot be fully replaced by AI.

Compliance strain

The survey also reports rising compliance pressure linked with AI in software pipelines. Many respondents said AI adds new complexity to regulatory and governance work.

Seventy-nine per cent said AI is making compliance harder. Eighty-five per cent reported that more compliance issues are now found after deployment.

Skills in this area are becoming more important for individuals. Thirty-seven per cent of respondents view AI-driven security and compliance as the top skill for career progression.

Expectations are shifting on how compliance will operate in future. Eighty-five per cent of those surveyed predict that by 2027, compliance checks will be built directly into code.

Toolchains under pressure

The survey findings highlight tension between faster coding and slower overall delivery. They show that toolchain complexity and governance requirements are key sources of friction.

GitLab said that fragmented toolchains are undermining the gains from AI-driven development. It also said that uncoordinated use of AI agents can worsen integration problems across teams.

"This survey illustrates what we call the 'AI Paradox,' where coding is faster than ever, yet the lack of quality, security, and speed across the software lifecycle is causing friction on the road to innovation," said Manav Khurana, chief product and marketing officer, GitLab. "Toolchain fragmentation has created bottlenecks for developers, and AI agents are amplifying the issue. Organisations need a new framework to match the speed of software development in the age of AI, one that provides intelligent orchestration across the entire software lifecycle while addressing the interconnected requirements of AI orchestration, governance, and compliance that individual point tools simply cannot solve."

The research suggests that Australian organisations will test new platform models and governance frameworks over the next few years as they seek better alignment between AI adoption, security and compliance.