Closing the AI execution gap: lessons from a year of transformation
This article was featured in our Ideas to Impact newsletter. Each month, we explore the ideas that create real impact with exclusive insights from our team to help accelerate your strategic business outcomes. Subscribe today and receive each issue in your inbox as soon as it’s published.
Looking back on 2025, one thing is clear: it was a revealing year for transformation. I watched hundreds of organizations wrestle with change, and I had countless conversations with our leadership team about what’s actually working and what keeps getting in the way. What follows are the uncomfortable truths we saw again and again: patterns that explain why so many AI initiatives stall despite enormous investment.
The uncomfortable truth about AI failures
After a year of relentless investment, pilots, and announcements, a hard truth has been hard to ignore: most organizations aren’t failing at AI because the technology isn’t ready. They’re failing because their ways of working aren’t.
Confidence is everywhere. Impact is not. Leaders speak fluently about transformation, yet struggle to point to sustained gains in speed, quality, or decision-making. The problem isn’t ambition or intelligence. It’s structural. AI can accomplish extraordinary things – but only within systems designed to adapt to change.
I’ve come to a simple conclusion: AI accelerates what already exists. That may sound promising until you consider the corollary. If the underlying ways of working are unclear, fragmented, or slow, acceleration simply magnifies the problem. If your ways of working are broken, AI won’t fix them.
Why technology alone won’t deliver results
This is the uncomfortable pattern playing out across industries as we head into 2026. Advanced technology is being layered onto operating models that were never designed for speed, autonomy, or continuous learning. Automation is expected to compensate for unclear priorities and diffused accountability. It rarely does.
The organizations seeing progress are making harder choices first. They’re simplifying how work flows, clarifying ownership, and removing friction – before scaling AI.
Yet belief continues to outpace capability. Our Global Intelligent Delusion survey findings point to a widening gap between perceived readiness and actual execution. “The delusion isn’t believing AI matters,” says Fredrik Hagstroem, our CTO. “It’s believing technology alone will deliver results.” Without changes to incentives, behaviors, and leadership habits, capability plateaus – regardless of how advanced the tools become.
Speed has only sharpened the contrast. Transformation programs are moving faster than ever, but many organizations are showing signs of strain. Initiatives stack up. Priorities shift. Teams absorb change without ever fully integrating it. “The future isn’t just faster,” says Anjana Mistry, our CFOO. “It’s more demanding.” What’s missing, she argues, are systems that protect focus and learning – not just momentum. Without them, urgency becomes the enemy of progress.
The AI execution gap: from knowing to doing
This helps explain why the conversation is shifting as 2026 begins. Insight is no longer scarce. Execution is. Knowing what needs to change is relatively easy. Making better decisions – repeatedly, consistently, in the flow of work – is not.
“Knowing isn’t the same as doing,” says Jaime Ruhl, our CRO, echoing a point many leaders quietly recognize. Transformation doesn’t stall because people don’t understand the strategy. “Real change,” she says, “happens when insight shows up in everyday decisions.”
But execution at scale introduces a new set of risks. As AI transitions from experimentation to daily operations, long-deferred questions surface quickly. Who is accountable? What behaviors are being reinforced? And what happens when speed outpaces judgment?
“AI should elevate human capability, not replace it,” argues Aldis Erglis, our CAIO. Designed well, he says, it sharpens decisions and strengthens learning. Designed poorly, it simply codifies existing flaws.
The real lesson: building transformation that lasts
That’s my biggest takeaway from 2025, and the challenge for 2026. AI will not rescue broken operating models. It will expose them, efficiently and at scale.
The organizations best positioned for what comes next won’t be the ones that moved fastest last year. They’ll be the ones willing to confront how work actually happens and redesign it accordingly. In a business culture obsessed with acceleration, building transformation that lasts may be the most radical move of all.
Additional insights
From AI strategy to impact: a product-led playbook
Thriving in the age of AI: the product-led advantage
Beyond the hype: turning AI into end-to-end business value
