What businesses are actually doing with AI.

Adoption is wide. Depth is not.

The U.S. Census Bureau's Business Trends and Outlook Survey found that between December 2025 and May 2026 about 17 to 20 percent of U.S. businesses reported using AI, with 20 to 23 percent expecting to within six months. Use climbs steeply with size: 37 percent of firms with 250 or more employees, 32 percent of those with 100 to 249, and under 20 percent of firms with four or fewer. Among firms with fewer than 20 employees, use was essentially flat over the period [1].

Sectors differ too. As of early May 2026, the Census reported AI use of 39.7 percent in the information sector and 33.9 percent in finance and insurance, against 14 percent in retail trade [1].

Broader measures run higher. The Stanford AI Index reported organizational adoption reaching 88 percent in 2025, and generative AI reaching 53 percent adoption among the population within three years, faster than the PC or the internet did [2]. The measures differ in how they count, so read the gap between them as a range, not a contradiction.

Agents are coming faster than the guardrails.

In Deloitte's State of AI in the Enterprise survey of 3,235 IT and business leaders in 24 countries, 74 percent expected their companies to use AI agents at least moderately by 2027, and 23 percent of those expected to use them extensively. Only 21 percent said their governance for agentic AI was mature, meaning clear limits on what an agent may decide, real-time monitoring and audit trails [3].

That gap is the reason we build approvals, limits and logs into every agent, and why our readiness checks look at governance as closely as data.

It comes down to people.

Microsoft and LinkedIn's 2026 Work Trend Index surveyed 20,000 workers across ten countries between February and April 2026. Sixty-five percent of AI users said they fear falling behind without rapid AI adoption, and 58 percent said they now produce work they could not have a year earlier [4].

The most useful finding is about where the impact comes from. Organizational factors such as culture, manager support and talent practices accounted for 67 percent of reported AI impact, against 32 percent for individual mindset and behavior. The most advanced users were far more likely to have managers who openly model AI use (85 percent versus 64 percent) and to pause before work to decide what a person should do and what AI should do (53 percent versus 33 percent) [4].

Where the evidence is strongest: clinical documentation.

Healthcare offers some of the most careful measurement. A study of about 1,800 clinicians at five academic medical centers found that ambient AI scribes saved roughly 16 minutes of documentation time and 13 minutes in the electronic record per eight hours of patient care, and that adopters saw about one additional patient every two weeks. STAT's headline called the savings modest, and use was inconsistent across clinicians [5] [6].

It is a useful reminder of what real results look like: a genuine gain that is smaller than the marketing, and larger for the people who actually use the tool.