Beneath the enthusiasm lies a harder truth: while appetite is high, actual readiness is low.

Across industries, organisations are realising something important - wanting AI is simple, but implementing it responsibly, consistently, and at scale is another matter entirely.

Propel Tech’s new practical guide for unlocking AI confirms what many leaders are discovering first-hand: the biggest blockers to AI adoption have little to do with AI itself. Instead, they stem from the systems, data and processes already embedded within the organisation. And unless those foundations are strengthened, even the most advanced tools will struggle to deliver meaningful results.

The real state of organisational infrastructure

There’s no doubt AI is transforming the workplace. PwC’s latest jobs barometer shows workers with AI expertise earning a 56% salary premium, while CEOs, including those at Anthropic and Ford, have openly acknowledged that AI could replace up to half of entry-level white-collar roles.

The urgency to adapt is real. Early movers almost always outperform the rest. 

Yet the accelerating pace of change has exposed a widening gap between ambition and operational capability. According to S&P Global Market Intelligence, AI project failure rates are surging, 42% of businesses discontinued the majority of their AI projects in 2025, more than twice the figure from the previous year.

And in most cases, the failures have nothing to do with shortcomings in AI technology. They come from the fundamentals: data quality, system compatibility, governance, and strategic clarity.

The first major barrier: Data

Data remains the biggest challenge. AI is only as good as the information it’s trained on. When data is duplicated across teams, mislabeled, stored in incompatible formats, trapped in legacy systems, or incomplete, AI simply cannot function effectively.

Propel Tech’s report highlights a clear trend: organisations with structured, connected data pipelines extract dramatically more value from AI than those relying on fragmented experimentation. Real-time, trustworthy data isn’t a bonus, it’s the prerequisite for AI-driven performance, automation and insight.

A common misconception? That AI will somehow fix broken data. In reality, the data has to be fixed first.

Older systems aren’t the problem

Many UK organisations still depend on older platforms not originally designed with AI in mind. While this can slow adoption, it doesn’t have to stop it.

A complete rebuild isn’t always necessary. Modern, bespoke software can integrate legacy infrastructure with newer architectures, enabling real-time data flow and the API-driven environment AI requires. AI transformation is often evolution, not revolution, as long as the underlying structure is intentionally modernised.

Strategy, ownership and governance still lag behind

Even when data and systems are strong, many organisations struggle due to a lack of clear AI strategy. The Institute of Directors’ recent analysis shows that half of UK leaders lack the skills to govern AI effectively, and many remain unsure how to evaluate or trust AI-generated outputs.

Andy Brown, director at Propel Tech, sums up the issue: 

“Many organisations start with pilots rather than outcomes. They adopt tools without defining ownership, monitoring, or even what ‘value’ means in an AI context. Without strategy and governance, AI becomes a series of disconnected experiments rather than a driver of meaningful change. That’s why so many projects stall.”

AI will reshape business - but only for those prepared to support it

AI is already reducing repetitive workload, enhancing customer experiences, streamlining operations and revealing new sources of insight. With the right foundations, AI becomes more than a tool, it becomes a competitive edge.

What’s becoming increasingly clear is that successful AI adoption isn’t powered by enthusiasm or experimentation, it’s powered by readiness. Organisations investing in modern systems, high-quality data, and clear strategic direction are the ones progressing. Those that don’t risk falling into the growing pool of failed AI attempts.

If you want AI to succeed, start with your foundations

Propel Tech’s full guide explores each of these themes in depth, from practical readiness checklists to cross-sector examples and avoidable missteps. For organisations unsure whether their systems, data or processes are truly prepared for AI, the report offers a clear and accessible starting point.

Download it for FREE here: BESPOKE SOFTWARE: THE KEY TO UNLOCKING AN AI-POWERED WORKFORCE

More reading

Legacy isn't a life sentence - Should you modernise or rebuild your software? 

Who's afraid of AI? - What different sectors really think of AI 

The resilient future of bespoke software - smarter, safer, more strategic