AI ambitions collide with legacy integration problems

Businesses are rapidly pushing artificial intelligence into their technology strategies, but many are hitting constraints caused by integration challenges and ageing systems, according to research commissioned by NashTech.

A survey of 1,000 technology decision-makers across EMEA, North America and Asia-Pacific found that 96% said AI is accelerating changes in technology strategy. Most also said AI is an immediate priority for custom software development, highlighting a gap between AI ambition and the technology foundations inside many organisations.

The findings suggest integration has shifted from an implementation concern to a risk issue. While 44% said they invest in custom software primarily to improve integration, 40% named it their biggest challenge. Another 47% said legacy integration could affect compliance.

John O'Brien, Chief Executive Officer at NashTech, said the limiting factor for many AI projects is the ability to connect systems and data flows, not the models themselves.

"The AI conversation has been dominated by models and use cases. But in most enterprises, the real constraint is not intelligence. It is integration," said John O'Brien, CEO, NashTech.

The report, Differentiating through custom software: The NashTech 2026 report on software development in the AI age, examines how custom software and commercial off-the-shelf products shape organisations' ability to scale AI across operations. It describes integration shortfalls as accumulated technical obligation that becomes harder to manage as organisations deploy AI across multiple business processes.

From pilots to rollout

Many enterprises have moved beyond experimentation and are preparing for formal deployment. The survey found that 85% have begun adopting AI or expect to do so within the next 12 months. Respondents also reported efforts to formalise AI governance, reflecting greater attention to risk, accountability and oversight.

Respondents also expect AI to change how they build and maintain software. Three-quarters anticipate a significant positive impact on custom software development, citing faster development cycles, automated testing, improved code quality and more adaptive, data-driven systems. These expectations come as many technology teams are still addressing fundamentals such as integration, data access and controls.

Integration sits at the centre of that tension. AI initiatives often depend on clean data, consistent definitions and reliable access across multiple applications, requirements that legacy estates can complicate. The survey links these constraints to compliance risks, including data retention, access controls and auditability across connected systems.

"AI does not fail because of algorithms. It fails because the systems beneath it were never built for intelligence at scale. What we are seeing is integration debt quietly turning into AI debt," said O'Brien.

Risk and governance

Security and privacy concerns featured prominently. Data privacy across systems was cited as a top risk by 49% of respondents, while 48% said they were concerned about third parties handling sensitive data. The results highlight the difficulty of managing information flows when AI systems interact with multiple internal applications and external providers.

Governance approaches varied. Fewer than half (47%) said board-level reporting forms part of risk management for AI and related technology work, suggesting uneven executive oversight as AI moves into operational settings where incidents can carry regulatory and reputational consequences.

The report also points to a sequencing problem: many organisations are moving towards more advanced generative and agentic AI systems while still building the groundwork around data management, integration and governance. For large organisations, these prerequisites can take years of architecture changes and process redesign, especially where estates include bespoke platforms and decades-old systems.

Leadership disconnect

The survey found a gap between senior leaders and mid-level managers on whether custom software projects meet expectations. While 63% of senior leaders said projects exceeded expectations, only 39% of mid-level managers agreed, suggesting those closer to delivery encounter more day-to-day friction.

Mid-level respondents cited two recurring issues: 36% identified scope creep, and 46% pointed to integration challenges. The findings suggest this perception gap can mask delivery strain until projects scale, particularly when executive confidence is not matched by operational readiness.

Quality over speed

Despite pressure to move quickly on AI initiatives, respondents said engineering quality remains a priority. While 46% aim to balance speed and quality, the survey suggests that when forced to choose, organisations tend to prioritise quality, particularly among senior decision-makers. The emphasis reflects caution around stability, security and maintainability, even as AI tools and methods promise faster delivery.

Partner expectations

The research also examined the role of external technology partners as organisations tackle integration and governance complexity. While 47% described their partners as trusted delivery providers, only 32% viewed them as truly strategic. Even so, 97% said they would invest in partners that consistently deliver long-term value.

NashTech operates 11 offices across nine countries and runs five delivery centres, with more than 2,000 engineers. It works across sectors including insurance, financial services, retail, logistics and higher education. The findings suggest service providers and internal teams will face growing demand for integration work and governance frameworks as AI moves from isolated pilots to broader operational deployment.