AI adoption and business results
Four studies on AI use, workflow redesign and the results companies report.
AI is already improving individual work, and some companies report gains in profit. The studies below show what helps turn that potential into a business result.
Personal productivity and business results
Two outcomes reported in McKinsey’s global AI survey.
McKinsey, 25 August 2026. The survey ran in May–June 2026 with 1,719 respondents in 97 countries. EBIT is earnings before interest and taxes.
The questions measure different outcomes.
For your business: AI can improve productivity, while its contribution to profit depends on how the saved time, extra capacity and costs affect the business. We assess those outcomes separately to decide where investment is justified.
Deployment and measured results
Share of S&P 500 companies making each type of AI disclosure in Q2 2026.
Apollo, 11 September 2026; figures also shown in a16z, page 27.
The categories overlap. They describe public reporting: a company can use AI without publishing a measure of its effect.
For your project: define the goal and how it will be measured before investing. An audit helps select a useful application and estimate its costs; a practical test checks whether it delivers the intended result.
Workflow redesign and reported value
Leaders at the initial adoption stage who report value from AI for their organisation.
McKinsey, 8 July 2026. The wider survey covered 750 AI users in February–April 2026; this comparison concerns leaders at the initial adoption stage.
The association supports examining workflow changes alongside tool adoption. A client’s expected benefit still needs its own assessment.
Why the process matters: buying a tool leaves decisions about inputs, review, responsibilities and the use of saved time. We help redesign those steps around AI and train employees to work with the changed process.
How enterprise customers use additional AI features
Share of weekly active users using each feature in OpenAI’s enterprise customer cohorts, June 2026.
Plugins combine reusable instructions with connections to data and tools. Skills provide reusable instructions for a task.
OpenAI Enterprise Signals, updated 12 August 2026; also shown in a16z, page 79.
The frontier cohort is the top 10% of companies by output tokens per active user; the typical cohort is the 45th–55th percentiles. This groups companies by use intensity.
How this guides implementation: choose the functions, connections and agents that fit the task, configure them and practise using them. We help move from trying individual tools to a workflow the team can use.
What this means for your project
We compare solutions using quality, total cost and changes to the process. A practical test helps decide whether to implement the selected option.
These charts were drawn by Veracto using published figures. The a16z State of Markets report, September 2026 was the starting point; the workflow and outcome comparisons come from McKinsey’s studies.
Write to us to discuss where AI might help. How we work explains what happens next.