According to recent research from Gartner, 77% of senior software engineers say that adding AI to existing applications is a "significant or moderate pain point." This pressure is largely driven by C-suite enthusiasm for AI's potential, especially agentic AI, which is currently dominating executive wish lists.

Many organisations remain limited by legacy systems which makes upgrades both delicate and expansive. A study from Pegasystems warns that these legacy dependencies don't just delay digital transformation, they also stall progress on broader AI strategies, putting companies at risk of falling behind competitors.

AI is becoming an increasingly useful business tool to combat legacy systems with the market for AI application development starting at $5.2 billion and expected to grow rapidly.

Both startups and major cloud providers are building platforms designed to simplify the AI upgrade process. Still, engineering leaders are urged to tread carefully in selecting the right tools.

Richard Bovey, Chief of Data at AND Digital, commented: "As AI adoption accelerates, many organisations prioritise rapid deployment, but often at the expense of a critical foundation, the skills of their workforce and the quality of their data. AI success doesn't come from tools alone, it depends on the people who use it and that they understand how to harness data effectively and apply to AI ethically.

"Without targeted investment in upskilling, companies risk building systems that are inefficient, biased, or poorly integrated as a result of fragmented data. According to our data loyalty research, 56% of business leaders are investing in AI without fixing their data problem first. In an increasingly competitive landscape, those who focus on upskilling their teams will be the ones that thrive, as they build a strong foundation for AI success."

Stuart Harvey, CEO of Datactics, commented: "AI is an extremely useful business tool, but the data foundations must be right to realise its full potential. Many enterprise environments still suffer from fragmented, siloed data and when engineer teams are asked to layer AI into systems with poor data, they face significant risk in inaccurate model outputs, unreliable responses, and integration complexity.

"In the rush to adopt AI, CEOs must first recognise the importance of data quality and readiness, as high-performing AI needs standardised, timely, and accessible data. Without that, application upgrades involving AI are not just inefficient, they're risky to deploy at scale."