The AI readiness gap: AI won't decide the next decade, rebuild speed will

Interrogating real-world case studies of firm’s that got AI right and those that didn’t reveal an uncomfortable pattern. The firms that fell behind rarely had worse technology; they had less readiness to use it and less appetite to change what needed to be changed. What we learned from a Harvard Business School AI Institute workshop with GHD leaders is that the readiness gap widens every quarter.

Industry data and research echo what surfaced at Harvard.

McKinsey's State of AI in 2025 finds that 88 percent of organisations now use AI in at least one business function, up from 78 percent a year earlier, yet only 39 percent report EBIT impact at the enterprise level. The single factor most strongly correlated with that impact is workflow redesign, not tool selection.

BCG's Widening AI Value Gap puts a sharper edge on it. Across 1,250 executives globally, just 5 percent of companies qualify as "future-built" and those firms are capturing 1.7x the revenue growth, 1.6x the EBIT margin and 3.6x the three-year total shareholder return of laggards. Sixty percent of the sample report minimal gains despite substantial investment. The case studies in the workshop at Harvard bore this out.

The better question isn't what AI can do. It's what your organisation is shaped to absorb.

Three levers define best practice. Data structure comes first: siloed, inconsistent data caps the ceiling of every downstream tool. MIT Technology Review Insights reports that only one in ten companies have scaled AI agents, with the delay traced not to model shortcomings but to data architectures that fail to deliver reliable business context. Most enterprise data is still shaped for reporting, not for reasoning.

Process design comes second. Automating a broken workflow gives you a faster broken workflow. The firms getting real value are redesigning processes around what AI makes newly possible, not inserting it into what already exists. Decision rights come third. Where the authority to redesign, not just deploy, is unclear, pilots proliferate and nothing compounds.

Almost none of this is a technology problem in disguise. Most of it is a leadership one.

What gave me confidence while workshopping at Harvard was watching our own leadership in the room. No one underestimated the scale of what is being asked of us. No one flinched from it either. Hard questions, honest self-assessment and a shared willingness to be changed by the answer.

Readiness is a leadership posture. The questions worth sitting with are uncomfortable ones. Where is our data still shaped for reporting rather than reasoning? Which processes are we automating when we should be redesigning them? What are we genuinely willing to be changed by?

Firms that answer honestly move faster. Firms that don't spend the next two years buying tools their organisations are not shaped to use.

The widening gap will not be closed by bigger AI budgets. It will be closed by coordination across functions, cadence measured in quarters rather than annual strategy cycles and the appetite to hold the discomfort of transformation without reaching for a smaller version of it. The advantage compounds for the fastest re-shaper. AI efforts demand pace and coordination. Simply, if you’re using AI, your organisation needs to be shaped to be rebuilt around it.