AI adoption surges but continuous use still lags badly

Most organisations have adopted AI tools, but only a small minority run them in a continuous operational model, according to research commissioned by Thoughtworks and conducted by IDC.

The study found that nearly 90% of organisations have integrated AI tools in some form. Only 12% said they had moved to what the report calls a fully continuous, AI-driven optimisation approach.

IDC surveyed 500 senior IT and digital decision-makers in the United States, the United Kingdom, Germany, and APJ. Participants were from organisations with more than 1,000 employees, across manufacturing, finance, retail, and life sciences. About 20% of respondents were based in the APAC region.

The findings highlight a gap between adoption and operational maturity. Many firms appear to treat AI and broader IT change as intermittent programmes rather than an ongoing operating model.

Workforce impact

The research also points to workforce pressures linked to project-based modernisation cycles, including attrition, burnout, and declining morale among staff involved in these efforts.

Across respondents, 17% said staff had left specifically due to the pressures of intermittent modernisation work. Respondents also linked burnout to reactive IT cycles, with 49% reporting burnout in the workforce. In addition, 45% of organisations reported a measurable decline in staff morale.

Thoughtworks framed the results as a warning about the sustainability of stop-start change programmes, as teams juggle platform upgrades, security patching, and rising demand for digital products.

Steven Yurisich, Regional Managing Director, APAC at Thoughtworks, said the findings point to retention risks in technology roles and operational strain from reactive cycles.

"The project mindset is broken. In Australia, the cost of reactive IT isn't just financial, it's a drain on talent and organisational momentum," said Steven Yurisich, Regional Managing Director, APAC, Thoughtworks. "Speed, resilience, and innovation won't come from bigger teams. They'll come from AI-enabled operating models built for continuous change."

Measured gains

Respondents who said their organisations had shifted from one-off initiatives to continuous modernisation reported better results across several metrics.

They reported 45% faster product and feature deployment, described as improved release velocity. They also cited a 48% reduction in risk exposure through AI-led security and vulnerability management, and a 36% improvement in system maintainability, which the research linked to scalability.

The report presents these as outcomes associated with greater maturity, but suggests many organisations remain in earlier phases. In those firms, modernisation appears to run as discrete initiatives, creating peaks of activity followed by periods of catch-up maintenance.

Operational shift

The report outlines a 180-day action plan, standardised by Thoughtworks, for organisations aiming to move from reactive maintenance to value-based operations. It groups the plan into technology execution, skills development, and contracting approaches.

On technology delivery, it points to "pipeline intelligence" and using AI for guided remediation and testing. The research suggests this could lead to more predictable change, with less reliance on manual interventions during release cycles.

On skills, the study identified AI and machine learning literacy as the most critical capability across the sectors surveyed. It describes an "upskilling mandate" to spread AI familiarity across delivery and operations teams, rather than limiting it to specialist groups.

Commercial models also featured in the findings. The study reported a shift towards shared-risk models, with 43% indicating a transition in that direction. It also found 56% pointing to contracts tied to innovation milestones.

For service providers and internal IT leaders, the contracting data suggests growing interest in engagement structures that link delivery to agreed measures, rather than time and materials alone. The report frames this as part of a broader move away from project-based delivery.

In APAC, where 20% of the sample was based, the results add to an ongoing debate about managing scarce technology skills amid continuing platform change and security demands. The report's attrition and burnout figures underline the risk that intermittent modernisation programmes can create sustained pressure without building durable operating practices.

The research focused on organisations that already use AI or plan to adopt it within the next two years. That scope suggests the adoption figure reflects a population already engaged with AI, while the maturity figure shows how few have embedded AI into continuous operations.

Thoughtworks expects more organisations to move from AI tool adoption to continuous operating models as modernisation becomes a permanent feature of IT and digital work.