The AI execution gap: What organisations get wrong before they start

A few years ago, we wrote about why digitisation so often failed — not because technology was lacking, but because organisations tried to innovate without changing how decisions were made, how teams were structured, and how work actually flowed.

What is striking about AI is not how different the pattern is, but how familiar.

AI is now firmly on the board agenda. Pilots are launched at speed. Generative tools are rolled out across functions. Investment continues to rise. And yet, for most organisations, meaningful impact remains elusive. MIT's NANDA initiative found that 95% of enterprise generative AI pilots fail to deliver measurable impact on P&L. 60% of companies are reaping hardly any material value despite substantial investment. And nearly two-thirds of organisations have not yet begun scaling AI across the enterprise at all.

These are not small numbers from fringe studies. They are mainstream findings from the most rigorous research organisations in business, and they point to the same conclusion we reached about digitisation a decade ago: this does not fail at the technical level. It fails at the organisational one.

AI is not a solution. It is a stress test

Many leaders treat AI as a shortcut – to productivity, to better decisions, or around organisational complexity that has been building for years.

It amplifies whatever already exists inside the organisation. Where structures are clear and decisions are well-governed, AI accelerates things. Where they are not – where ownership is fragmented, where decisions stall, where responsibility is distributed by role rather than by capability – AI makes that friction more visible, faster, and harder to ignore.

This is not a failure of ambition, but a failure of readiness.

Pilots work because they are insulated from the organisation's structural problems: a small, motivated team, a focused use case, clear ownership. Then the pilot ends, the project team disperses, and the output is handed over to a broader organisation that was never redesigned to receive it. Nothing around the tool changes – workflows, decision rights, incentives all remain exactly as before.

The organisation stays the same while the technology moves on. AI cannot compensate for that imbalance.

The constraint nobody talks about: organisational legibility

In our work at Up Strategy Lab, this shows up with striking consistency.

The most consistent finding across AI transformation research is not that companies lack ambition, budget, or access to good models. It is that most organisations cannot clearly see how they actually operate. Work flows cut across teams in informal, undocumented ways. Decisions are delayed by invisible dependencies. Capabilities are assumed rather than verified. Skills are treated as fixed attributes attached to job titles rather than as a dynamic picture of what the organisation can actually do.

The decisive blockers are not technical: they are unclear ownership, legacy processes, skills gaps, and weak alignment between strategy and execution.

AI systems require clarity to function well – clear inputs, clear ownership, clear integration into real work, clear accountability for outcomes. Without that organisational legibility, AI stays trapped in experimentation regardless of how sophisticated the model is.

When speed hides inertia

AI adoption can look fast while moving nowhere.

Tools get deployed, pilots launch, dashboards appear – all within weeks. But beneath that surface velocity, decision-making stays slow, ownership fragments, and escalations persist exactly as before. What changes is the technology. What does not change is how the organisation actually moves.

Applying AI to existing processes without redesigning how decisions are made, who has authority to act, and how work flows across teams produces incremental efficiency at best – and expensive confusion at worst. Without aligned incentives, redesigned decision processes, and an AI-ready culture, even the most advanced pilots won't become durable capabilities. The real gains appear when organisations stop treating AI as an overlay and start rethinking how work flows across functions, not just within them.

Skills as execution infrastructure

The most underrated dimension of AI readiness is skills clarity – and it is also our favourite subject.

Skills matter here not because AI requires more training programmes, but because skills are how organisations understand what they can realistically change. Without a clear, shared view of capabilities, leaders cannot determine where AI should redesign workflows, shift decision rights, or redistribute responsibility – or identify which teams can absorb new ways of working and where gaps will stall progress.

When skills are treated as static role attributes rather than as a dynamic view of organisational capacity, AI has nowhere to land. It becomes another layer added to an already opaque system, rather than a lever for redesigning how work is done.

It is a pattern Up Strategy Lab encountered repeatedly in client work: organisations that could not see their own capabilities clearly could not make good decisions about where to apply new technology. That problem led Noel Braganza, co-founder of Up Strategy Lab, to build MuchSkills – not as a product in search of a problem, but as infrastructure for exactly this challenge. The platform grew from a colour-coded Google Sheets prototype to a globally recognised skills intelligence platform, winning the Red Dot Design Award and recognised as a Major Contender in the Everest Group PEAK Matrix® 2026.

What the organisations that get it right actually do

Organisations that succeed with AI share a common trait – they do not start with tools, they start with making the organisation legible to itself.

Before scaling AI, they understand how work actually flows across teams, where decisions are genuinely made, and how responsibility is distributed in practice. They develop a clear view of their capabilities – not as static roles or titles, but as the skills that determine what the organisation can realistically execute. Teams are designed around outcomes, and incentives are adjusted so that AI-enabled ways of working are reinforced rather than quietly resisted.

Most organisations are still struggling to generate meaningful returns from their AI initiatives – not because the technology is lacking, but because the organisational conditions to absorb it are not in place.

As Sylvain Duranton, Global Leader of BCG X, put it directly: "Companies cannot simply roll out GenAI tools and expect transformation. The real returns come when businesses invest in upskilling their people, redesign how work gets done, and align leadership around AI strategy."

This is not glamorous work. It does not generate impressive pilot results or board-level headlines. But it determines whether AI creates sustained value or sustained friction.

The question worth asking before your next initiative

It is tempting to frame all of this as an AI maturity issue – to put the problem out in front of you, somewhere you will get to eventually. That would be convenient, and incorrect.

The obstacles surfacing in AI adoption did not arrive with AI. Siloed structures, slow governance, role-based staffing, limited skills visibility, and weak links between strategy and execution have shaped organisational performance for years. AI simply removes the buffer and compresses the gap between intent and reality.

The most important question before your next initiative is not: what AI should we deploy? It is: are we structurally ready to use it?

That readiness is not defined by your vendor, your model, or your pilot results. It is defined by clarity – about capabilities, about decision-making, about how work is coordinated and how accountability is held.

AI rewards organisations that understand themselves. It exposes those that do not.

If answering that question feels difficult internally, that difficulty itself is a signal worth paying attention to.

Frequently asked questions

Why do AI pilots succeed but fail to scale? 

Pilots work because they are insulated from the broader organisation's structural problems – small team, clear scope, focused ownership. Scaling fails when outputs need to be absorbed by workflows, decision structures, and incentive systems that were never redesigned to receive them. The technology moves forward; the organisation stays the same.

What is the strategy-execution gap in AI? 

The strategy-execution gap in AI refers to the disconnect between what organisations plan to do with AI and what actually changes in day-to-day operations. AI exposes this misalignment between strategy and execution faster than any previous technology – making visible the fragmented decision rights, legacy workflows, unclear ownership, and skills gaps that were always there but easier to ignore.

How can organisations close the gap between AI strategy and execution? 

Closing the gap requires organisational work before and alongside technology deployment: clarifying how decisions are actually made, mapping how work flows across teams, developing a dynamic view of capabilities rather than a static skills inventory, and redesigning workflows end to end rather than layering AI onto existing processes.

What role does skills visibility play in AI readiness? 

Skills visibility is a prerequisite for AI readiness, not a nice-to-have. Without a clear, current view of organisational capabilities, leaders cannot determine where AI creates leverage, which teams can absorb change, or where gaps will stall progress. Skills data is execution infrastructure – the foundation that makes strategic decisions about AI actionable.

If this is the constraint your organisation is hitting

This is the space Up Strategy Lab works in – where organisational structure determines whether strategy can actually be executed.

No tool or technology can compensate for organisational opacity. We start by making your organisation legible to itself: mapping how work actually flows, where decisions genuinely stall, and where AI creates the most meaningful leverage. That clarity comes first – because tools built on top of an opaque organisation just make the opacity more expensive. Once we understand how your business actually operates, we build and deploy working tools that fit those realities – prototypes you can test with your team, solutions that get embedded into how work actually gets done.

If your AI initiative has stalled – or if you are trying to work out where to start – let's talk.