Insights on AI adoption, decision intelligence, and digital transformation
Most organisations pick their AI initiatives, then run them. The sequencing follows executive priority or budget cycles, not dependencies. When dependent initiatives launch before their foundations exist, they stall or build workarounds that become technical debt.
Organisations focused on fewer AI initiatives see 2.1× better ROI. But the standard advice (focus on three to four use cases) applies the principle at the wrong altitude. The mechanism works at the domain level, where each function chooses initiatives based on friction only they can see.
AI initiatives get scoped in isolation, but their consequences propagate through every connected process, team, and system. Two in three companies struggle to reimagine workflows for AI, because the cascade reaches people, not just processes.
78% of organisations use AI in at least one function. Only 1% describe their implementations as mature. The gap is context: AI works at the edges because endpoints don't need organisational understanding. Moving closer to the core requires value chain engineering first.
One in four failed AI initiatives traces back to weak governance. In most cases, ownership was vague enough that everyone assumed someone else was handling it. Accountability means someone specific owns every handover point, with defined scope, clear triggers, and an escalation path.
The knowledge that determines whether an AI initiative will work in production is distributed across the organisation. Most of it can't be extracted by asking.
65% of high-performing organisations define when AI outputs need human validation or intervention, compared to 23% of everyone else. What separates them is alignment: business, operations, and technology sharing an understanding of what the system does, when it'll be wrong, and what happens in those situations.
Sponsoring a transformation and setting its direction need different knowledge. When leadership imposes direction without discovery, even fully backed initiatives fail.
Only 12% of organisations report sufficient data quality for AI. But data quality is the surface issue. The harder problem is that the same data means different things depending on who's using it and why.
42% of companies abandoned most AI initiatives in 2025, up from 17% in 2024. The helpers organisations bring in — consultants, vendors, integrators — all hit the same wall. The knowledge needed for good AI decisions is distributed across the workforce, and none of them can access it.
Only 21% of organisations have governance mature enough for AI agents, while agentic AI usage surges from 23% to 74% within two years. The knowledge you need is usually there — it's just scattered across parts of the organisation that don't talk to each other.
62% of organisations are stuck running AI pilots that never reach production. Only 7% have fully scaled. The gap between pilot and scale isn't technology; it's that each transition demands organisational elements that pilots never tested.
53% of CEOs say their teams can't align on their AI priorities. The root cause is structural: strategy, capability, and operationalization live in different parts of the organization, and in most companies, they never meet.
Harvard Business Review reports 80% of enterprise AI projects fail. Traditional IT had ~42% failure rates; add AI to the mix and they double. The real obstacles are organizational, not technical.