United States
Get AI working in a US business.
Your work mapped, the rules that keep people in the decisions that need them, the system built on top, and managers who can supervise it without us in the room. Working across US time zones from Canada.
Where adoption sits
A fifth of US businesses are using AI
The national rate has held between 17 and 20 percent since late 2025. Underneath it the spread by company size is wide enough to be the story on its own.
19.8%
of US businesses reported using AI as of 3 May 2026, on a national basis.
37%
of firms with 250 or more employees, against under 20% at firms with four or fewer.
39.7%
in the Information sector, the highest of any, with finance and insurance at 33.9%.
20-23%
of businesses expected to be using AI within six months of being asked.
US Census Bureau, Business Trends and Outlook Survey, covering 14 December 2025 to 3 May 2026. Read the release. Firms with 100 to 249 employees came in at 32%, and firms with 20 or more employees were the only group whose adoption rose over the period.
What stops it
Four reasons US projects stall
Each one has an answer that comes before the build. Three of the four are decided in the first two weeks of an engagement.
-
The gap is size, not sector
A 250 person firm is roughly twice as likely to be using AI as a 4 person one. The difference is rarely the technology and is usually that somebody at the larger firm was given the job of working out where it fits. Smaller firms get the same answer from a map, without the headcount.
Process mapping and roadmap -
Fifty states, different answers
There is no federal AI statute. The rules that bind a US business come from states, they were written on different legal theories, and they take effect on different dates. A policy written for one state is a starting point everywhere else.
AI governance setup -
Tools bought before the work was mapped
The common US pattern is a licence per department and no system underneath. Each one works. None of them talks to the others, and the reporting still gets assembled by hand at month end.
Systems and AI build -
No way to tell whether it worked
Adoption gets reported as seats issued and logins counted. Neither is a result. A number measured from a baseline taken before go-live is the only version of this that survives a board meeting.
AI operating system
What already applies
The rules that govern AI in the United States today
A business operating in more than one state is inside more than one regime. The obligations come from state statutes and privacy regulations that are being revised while organizations are still reading them.
- There is no federal AI statute. State law is the operative law of the United States on AI. Every state introduced AI legislation in 2025, and they diverge on the most basic question of what the law should regulate.
- Colorado rewrote its own law in 2026. Senate Bill 26-189 was signed on 14 May 2026 and takes effect on 1 January 2027. It repeals the 2024 Colorado AI Act and replaces it with a disclosure framework built around automated decision-making technology in consequential decisions. Summaries describing a duty of care and impact assessments are describing the repealed version.
- California regulates automated decisions through privacy law. The CCPA regulations carry risk assessment and automated decision-making technology requirements, which reach any business meeting the thresholds.
- Employment is regulated separately again. Illinois, New York City and Connecticut each impose their own obligations on AI used in hiring and employment decisions.
- Which law applies follows the person, not your address. A hiring tool that screens an applicant in Illinois, or an automated decision affecting a Colorado resident, sits inside that state regime wherever your company and your servers are.
Current as of August 2026. This is a summary of the landscape and it is not legal advice. Where a decision turns on it, take advice from counsel qualified in the relevant state.
How the work runs
Start with the process, then decide on AI
Digital adoption is the whole job and AI is one route through it. Most US engagements start at the map, because the cheapest way to find out whether AI belongs in a process is to price the process first.
Start here
Find out where you stand
Twelve questions across four dimensions, answered in about three minutes. You get a score out of 48, the tier it puts you in, and the one thing worth fixing first.