60% of organisations deploy untested code as AI accelerates development
Six in ten organisations are releasing untested code into production, according to research from Tricentis. The survey of 2,501 senior technology and engineering professionals across the US, UK, Ireland, Germany, Japan and Singapore reveals a shift in why this happens: teams are now making deliberate trade-offs rather than stumbling into quality failures.
Leadership pressure to prioritise speed accounts for 32% of cases. Another 30% cite the sheer volume of AI-generated code as too large for teams to test fully before deployment.
The pattern is consistent across sectors. Financial services leads at 64%, followed by retail at 63% and energy and utilities at 58%.
The confidence gap
A stark divide exists between boardroom and technical staff. Some 81% of CEOs express high confidence in AI-driven systems, compared with just 56% of QA and DevOps professionals.
The gap widens on preparedness to govern AI agents through the development lifecycle. While 44% of C-level executives say their organisation is very prepared, only 23% of QA and DevOps professionals agree.
Nearly half of organisations have fully implemented AI internally. Of those, more than half report their AI tools and processes change regularly.
What's slowing teams down
One-third of teams identify tool complexity and sprawl as a major barrier to continuous software quality. Skills gaps affect 33%, while 28% say code volume is growing faster than they can manage.
Security concerns trouble 27% of respondents. Another 24% each cite skills shortages and data quality problems.
Despite high stated confidence in agentic AI - 83% trust it to make release decisions - day-to-day operations reveal persistent obstacles. The research suggests confidence on paper doesn't match reality in practice.
The financial cost
One in five organisations lose between $1 million and $5 million annually to poor software quality. A further 45% estimate losses between $500,000 and $1 million per year.
Security and compliance failures account for 30% of those losses. Technical debt and rework contribute 28%.
Kevin Thompson, CEO of Tricentis, said the combination of faster development and weaker controls has become unmanageable. "With increased speed comes increased risk," he said. "When software quality processes fail to keep pace with development speed, organisations often respond by taking shortcuts that materially degrade confidence."
Thompson added that many organisations rely on quality processes designed for earlier stages of software development. "As development accelerates, leaders need clearer visibility into software quality risk and stronger alignment between engineering, QA and the broader business. The organisations that succeed will be the ones that can scale speed and control together," he said.
For development teams managing AI-generated code, understanding how to maintain quality at pace is becoming critical. AI for Software Developers offers structured guidance on integrating AI tools while maintaining standards.
Your membership also unlocks:
