Ometz AI

Research · 3 min read · September 2, 2026

State of AI adoption, September 2026: what the numbers mean for an owner-led business

Adoption is universal, scaling is rare, and the measured gains sit in a few specific kinds of work. A monthly read of the published research, translated for a business with a front desk and a back office rather than a data science team.

This is the first of a monthly read of the adoption research, kept deliberately short and confined to published sources. The purpose is practical: an owner deciding where to spend a limited budget should know what the evidence actually says, and what it does not.

Adoption: everyone

McKinsey's State of AI survey, fielded in mid-2025, found 88% of organisations using AI in at least one business function, up ten points in a year. Stanford's AI Index 2026 puts the money behind it: $582 billion of corporate AI investment in 2025, up 130% in a single year. If a competitor has not tried AI somewhere, they are now the unusual one.

Scaling: almost nobody

The same McKinsey survey found only 7% of organisations had fully scaled AI across the business. The majority are experimenting or piloting. For an owner-led business this is the most useful number in the research: the field is wide open, because the enterprises with the budgets are stuck in committee.

Where the gains are measured

Stanford's Index reports documented productivity gains of 14 to 26 percent in customer support and software development, the two functions with high-volume, structured, countable work. Harvard Business School's 776-person field experiment at Procter & Gamble found an individual working with AI matched the output of a two-person team without it. Both findings share a shape: repetitive work, high volume, an outcome you can count.

What that means without a data science team

You have those functions. The phone, the follow-up queue, intake paperwork, scheduling, reporting. You do not have a pilot committee for a project to die in. The research says pick one workflow with volume, baseline it, deploy against it, and measure. The 88% bought tools. The 7% changed an operating rhythm and gave it an owner.

What to watch next month

Whether agent deployment moves beyond pilots in the next survey wave, and whether the productivity evidence broadens beyond support and software. This read is updated when the sources are, and only then.

Sources

Want the next analysis first?

Keep reading

Next step

Find out where your operation leaks.

Fourteen questions, four minutes, and a maturity radar across the seven dimensions of AI readiness, from strategy to culture. Then decide if a conversation is worth your time.