The Curve · From the Evidence Register · Updated monthly · Next update 6 July 2026

The gap between using AI and creating value with it.

Three lines tell the story of the decade. Adoption runs fastest in the history of general-purpose technology. Production use trails it by a wide margin. Measured value trails further still. AiR Bench measures that distance, then helps you close it.

02550751002023202420252026ChatGPT momentAgentic wave beginsGenAI Divide publishedAiR v1.0
Using AI somewhereProduction use casesMeasured value% of organisations · Sourced from the Evidence Register · Updated monthly

01

Adoption

The share of organisations using AI in at least one function. This line runs fastest in the history of general-purpose technology.

02

Production

The share redesigning a workflow end to end around AI. It trails adoption by a wide margin, and marks where value research points.

03

Measured value

The share showing measurable impact against a baseline. It trails further still, and it is the line nobody else plots.

Three numbers, one distance

88%

of organisations now use AI in at least one function. Adoption no longer sets you apart. It is the weather.

McKinsey, The State of AI 2025

21%

have redesigned even one workflow end to end around AI. This is the production line's ceiling, and where the value research points.

McKinsey, The State of AI 2025

5%

of enterprise AI pilots delivered measurable P&L impact. The measured-value line is the one that matters, and the one nobody else plots.

MIT NANDA, The GenAI Divide 2025 · 300 deployments analysed

What joins The Curve next

Aggregate AiR data, as the dataset matures.

Once sector cells clear the minimum sample of 25, the AiR dataset joins this chart: measured readiness and realisation for the UK mid-market, by sector, over time, a line nobody else holds. We annotate every instrument version here, so you can see what evolved, when, and what evidence moved it.

Add your data point to The Curve

AiR Bench

Benchmark your business, and become one of the lines.

Add your data point →