UK enterprises move AI projects into production

OutSystems has published research showing that 91% of UK enterprises have moved AI projects into production. The survey also found that 96% of organisations globally are using AI agents in some form.

The findings point to broad adoption of AI agents among large organisations, but also suggest many companies are struggling to manage them consistently. According to the survey, 94% of organisations are concerned that AI sprawl is increasing complexity, technical debt and security risk, yet only 12% have implemented a centralised platform.

The report is based on a survey of nearly 1,900 global IT leaders and examines adoption levels, development approaches and governance practices as companies move from pilot programmes to operational use.

UK Barriers

In the UK, integration with existing systems was the main obstacle to AI success, cited by 43% of respondents. That was followed by the impact of legacy systems or fragmented data sources at 42%, and governance or compliance concerns at 38%.

A lack of internal skills was also a notable constraint for British organisations. Some 31% of UK respondents identified this as a barrier, the highest level among surveyed countries apart from Japan.

While 91% of UK enterprises said they had successfully moved AI projects into production, only 41% said more than half of those projects were successful. UK businesses generally rated their own AI maturity as intermediate rather than advanced.

That pattern was echoed in several other markets, including Australia, Brazil, Germany, the Netherlands and the US, while France was described as being at an earlier stage. Financial services and technology groups reported the highest levels of production deployment.

Agent Growth

Across all markets, 97% of organisations said they were exploring system-wide agentic AI strategies. Nearly half, or 49%, described their agentic AI work as advanced or expert.

The data also showed a fragmented technology landscape. Globally, 38% of organisations said they were combining custom-built and pre-built agents, a model that can make systems harder to standardise and secure. Most enterprises are still using governance approaches that vary by team and region.

AI use appears strongest in IT and software development, where businesses can track outcomes more directly. Thirty-one percent of global respondents said AI is already integral to their development practices, while another 42% said it has been embedded into specific phases of the software development lifecycle.

In the UK, 65% of respondents said generative AI-assisted development was now their main app development process. That compared with 60% who said they were customising SaaS applications internally and 58% who relied on outsourced or vendor-built solutions.

The survey also found that 52% of organisations globally now rely on a human-in-the-loop model, allowing systems to operate with reduced direct oversight while retaining supervisory control.

Scott Finkle, Vice President of Technology at McConkey Auction Group, described how the company approached an early AI deployment. “Our approach to working with OutSystems for an agentic solution was to start with a small, well-defined project that we felt we could get into production and that would actually have an impact on the business,” he said. “Our main goal was to build some muscle for developing AI projects moving forward. OutSystems and Agent Workbench will pay great dividends to us as we iterate on our AI implementation.”

OutSystems used the findings to argue that the challenge for many enterprises is shifting from adoption to control and architecture. The company recently introduced what it calls Agentic Systems Engineering, which it described as an open approach to building and managing governed agentic systems.

Chief executive Woodson Martin said the market had moved beyond experimentation. “The transition from AI experimentation to measurable business outcomes is no longer a future state-it is our current reality. The findings in the State of AI Development Report reveal a fundamental shift where building software and building AI systems have become one and the same,” he said. “As organisations move toward a 'system of agents' model, the challenge is no longer just about adoption, but about creating a stable architectural foundation that can coordinate these complex intelligent systems to drive real-world productivity.”