As organisations increase their investment in AI, capability building is often discussed as a secondary concern.

The strategy is approved.
The technology is selected.
Training is planned later. Staff are involved too late.

This sequencing creates risk. AI capability gaps are one of the most common reasons strategy execution slows or stalls, even when intent and investment are strong.

 

Why AI capability gaps show up during execution

AI changes how work is done, not just the tools people use.

It affects:
• How decisions are made
• How problems are diagnosed
• How work is prioritised
• How accountability is shared

When people do not have the skills to operate confidently in this environment, execution suffers. Decisions are deferred, reliance on specialists increases, and AI-enabled processes are underused.

Recent research reinforces this challenge. The World Economic Forum’s Future of Jobs research shows that 39% of core workplace skills are expected to be obsolete or require significant change by 2030 (WEF, 2025). This creates pressure on organisations to reskill but it is not happening fast enough.

 

Why training alone is not enough

Many organisations respond to capability gaps by rolling out training programs.

Training is important, but it is rarely sufficient on its own.

AI capability is not only technical. It includes:
• Understanding how AI supports decisions
• Knowing when to trust outputs and when to challenge them
• Working effectively alongside AI-enabled workflows
• Applying judgement in ambiguous situations

McKinsey’s 2025 State of AI research indicates that organisations investing in AI training without integrating learning into day-to-day execution struggle to realise sustained value from AI initiatives (McKinsey, 2025).

Capability must be built in context, not in isolation.

 

Capability building is an execution design issue

I see capability gaps emerging during AI execution. For example, at one multinational client, leaders are insufficiently upskilled on using AI in their day to day work. This means that as they rollout workflow changes to administration and marketing teams, the staff are resistant to AI changes as they think it is only them being having to work in new ways. Importantly for staff, the training on AI has been left too late during the implementation process so the learning curve is steeper.

A common issue is also expecting people to adapt informally and adopt AI without formally adjusting workflows and providing training in new ways of working. As a result, workflows become quickly non-standardised and haphazard.

Those who figure out through trial and error how to do their work with AI more efficiently are not identified, acknowledged or encouraged. This is a lost opportunity to share the best, most creative ideas. This type of “shadow AI” is very common in organisations. HBR quotes Ivanti’s study (2025) that 32% of workers are keeping their AI use hidden from their employees.

Effective organisations design execution in a way that:
• Embeds learning into real work
• Creates safe opportunities to experiment
• Clarifies expectations around new skills
• Reinforces capability through governance and feedback

This approach accelerates both confidence and performance.

 

Early involvement supports faster capability development

Capability building is more effective when people are involved early.

When teams participate in shaping AI use cases and workflows, learning starts before formal training begins. People understand why new skills matter and how they will be applied.

Gartner research has shown that early involvement in change initiatives increases willingness to adopt new ways of working and improves confidence in navigating change, particularly in technology-driven transformations (Gartner, 2023). This supports many other research studies from Change Management, Human Centered Design and Open Strategy disciplines – when people are involved in change early, they are more supportive and change is more successful.

In AI initiatives, this early learning reduces execution friction later. It is a key part of addressing the fear that people feel due to AI.

 

Leadership capability matters as much as technical skill

AI capability gaps are not limited to frontline teams.

Leaders also face new demands:
• Interpreting AI-supported insights
• Making decisions in more dynamic environments
• Setting appropriate guardrails
• Role modelling learning and adaptability

MIT Sloan research highlights that leadership capability is a critical factor in effective AI adoption, particularly where leaders are required to balance automation with judgement (MIT Sloan, 2025).

Without leadership capability, execution loses direction.

 

Integrating capability into strategy execution

Organisations that execute AI strategies well treat capability building as part of execution, not a separate activity.

They:
• Identify critical skills during strategy design
• Align capability development with implementation milestones
• Reinforce learning through governance and performance management
• Review and adjust skills as strategies evolve

This integration encourages adoption and accelerates sustainable outcomes.

 

From skills development to execution strength

AI amplifies both strengths and weaknesses in organisational capability.

When skills are aligned with strategy execution, AI accelerates performance. When gaps persist, AI initiatives remain under utilised and adoption languishes.

Capability building is not an HR exercise. It is a strategic execution discipline.

Organisations that recognise this early are far better positioned to turn AI ambition into consistent results.

PS More Ways to Accelerate your Strategy

Looking to move your strategy forward? Lisa Carlin works with selected clients directly to go from strategy to plan to result. Email Lisa.

A bespoke in-house Bootcamp is the fastest way to learn. Get the skills your teams need to go from Strategy to Plan to Results. AI optional. Bootcamp info here.

Subscribe here to ensure you don’t miss this weekly newsletter.

Connect with Lisa on LinkedIn.