Skar Education · United States
Essentials of
AI adoption.
Understand the shift. Find the business value.
A guide to how American businesses are using artificial intelligence, where investment is flowing, and how to turn new capabilities into useful work.
01 / The essentials
A business capability.
A learning curve.
Artificial intelligence uses computational models to recognize patterns, make predictions, or generate outputs. In a business, that can mean extracting information from documents, forecasting demand, or helping an employee draft a response.
Adoption means putting that capability into a working process—with clear inputs, accountable people, and a way to evaluate the result.
Understand the task
Generative AI produces text, images, or code. Other AI systems classify, predict, or detect patterns. Match the capability to the work.
Understand the evidence
A business using AI is not necessarily using it across every team. Adoption, investment, and productivity measure different things.
Understand the economics
Value comes from a better result or a lower total cost. Include review, integration, training, and mistakes when assessing a tool.
02 / U.S. adoption
AI is entering everyday business.
The Census Bureau’s Business Trends and Outlook Survey provides a national view of employer businesses. Its October 22, 2026 release shows how many reported using AI and how many expected to use it in the next six months.[1]
Reported current use
AI used in at least one business function during the two-week reference period.
Expected use
Businesses expecting to use AI in the next six months. An intention, not a realized outcome.
Read the denominator
These are shares of U.S. employer businesses—not employees, revenue, or all Americans.
Reference period: September 21–October 4, 2026. Estimates are subject to sampling error. See source notes for definitions.
Adoption by sector
The starting point
is different in every industry.
Information and professional services show higher reported use than several sectors centered on physical operations. For an owner, the useful question is which part of the business can benefit—not whether the company resembles a technology firm.
Source: U.S. Census Bureau, BTOS, Sector workbook, Q7 “Yes,” period 202621.[1] Selected sectors, not a complete ranking. Bars share a zero-to-100% scale.
View figures and sampling uncertainty
| Sector | Using AI | Standard error (percentage points) |
|---|---|---|
| Information | 46.4% | 2.02 |
| Professional, scientific & technical services | 43.8% | 0.93 |
| Finance & insurance | 40.5% | 1.53 |
| Manufacturing | 22.9% | 1.35 |
| Retail trade | 18.5% | 0.73 |
| Construction | 16.1% | 0.79 |
| Accommodation & food services | 10.7% | 1.08 |
Standard errors describe sampling uncertainty, not all potential survey error. Differences shown here have not been tested for statistical significance.
03 / The AI economy
Capital is building
new capabilities.
Stanford’s 2026 AI Index reports $285.88 billion in U.S. private AI investment in 2025. That is capital invested in AI companies, not the size of the AI economy or a measure of customer savings.[2]
For business owners, the practical implication is a changing supply of tools and services. Revisit what is possible, but evaluate each product against your own costs, customers, and operating requirements.
U.S. private AI investment / 2025
$285.9BRounded from $285.88 billion · Stanford AI Index 2026
Investment is an input.
Business value is the test.
Funding can support research, infrastructure, and new products. It does not establish that a particular tool is reliable, profitable, or right for your company.
How to evaluate a first application ↓04 / Business value
Why learning to use AI matters.
Building experience helps a business judge new tools, train its people, and identify work that can improve. The advantage comes from knowing where AI works, where it fails, and how to use the time or capacity it releases.
Evidence in practice
14%More issues resolved per hour
A customer-support study
A 2023 NBER working paper studied an AI assistant introduced to 5,179 customer-support agents. Access to the assistant increased issues resolved per hour by 14% on average, with larger gains among less experienced workers.[3]
One company and one workflow; this is not a forecast for all businesses. Results are from the November 2023 working-paper version.
The lesson: test a defined task with a measurable outcome. Do not assume the same improvement in a different setting.
Where to look in your business
| Workflow | Possible use | Measure and review |
|---|---|---|
| Customer service | Draft replies from approved product and policy information. | Handling time, factual accuracy, and escalations; a person approves replies. |
| Administration | Extract fields from invoices or summarize incoming documents. | Correct fields, exceptions, and review time; verify before updating records. |
| Sales & marketing | Prepare first drafts and organize customer research. | Editing time and qualified responses; verify claims and permissions. |
| Operations | Explore demand forecasts or flag unusual patterns. | Compare against the existing baseline; an owner checks business consequences. |
05 / Getting started
Start small.
Measure the whole job.
A useful first application has repeated work, accessible inputs, and someone who can judge quality. Choose a task where you can compare the current process with an assisted version before expanding it.
Choose one workflow
Record volume, time per task, current quality, and the cost of a mistake. Name the person responsible.
Set the boundaries
Use approved information. Check the tool’s data handling and access settings. Define what requires human approval.
Test real cases
Include ordinary work and difficult exceptions. Measure total time, including review and correction, against the same baseline.
Decide from results
Continue, revise, or stop. Expand only when quality and economics hold up in everyday use.
Your first exercise: define a pilot
- Today we spend ___ minutes on ___, ___ times each month.
- AI will assist with ___; a person will approve ___.
- We will test ___ representative cases, including ___ difficult cases.
- We will continue only if quality meets ___ and total handling time falls to ___.
- ___ owns the pilot and will review the results on ___.
Common questions
Before you begin.
Is AI adoption the same as automation?
No. Automation follows a process with less manual intervention. AI can help interpret, predict, or generate information inside that process. Some problems are better served by a straightforward rule, a template, or a clearer handoff.
Does every business need its own AI model?
No. A first pilot can use an existing product. Start by evaluating its performance, permissions, data handling, integration, and cost. Custom development should follow a demonstrated need.
Why do different reports show different adoption rates?
They ask different questions of different populations. A survey of large-company executives is not the same as a representative survey of employer businesses. Check geography, date, business size, and whether “use” means experimentation, production, or any business function.
When should a business expand a pilot?
When it meets defined quality and cost targets on representative work, staff know how to handle exceptions, and the owner can monitor performance. A convincing demonstration alone is not enough.
Sources & measurement notes
Know what the numbers describe.
- U.S. Census Bureau — Business Trends and Outlook Survey. Survey overview; National workbook; Sector workbook. Period 202621, released October 22, 2026; reference period September 21–October 4. Q7 “Yes” measures current use; Q24 “Yes” measures expected use in the next six months. National standard errors: 0.27 and 0.21 percentage points, respectively. Employer-business estimates do not include nonemployer businesses. BTOS excludes farms. The downloaded workbooks currently label the technology “Super Intelligence or SI (formerly referred to as Artificial Intelligence or AI)”; this page uses AI for continuity. Exact question text is preserved in the data file. Census broadened the question to any business function on November 17, 2025; comparisons with earlier production-only figures are not like-for-like. Definition change explained by Census.
- Stanford HAI — 2026 AI Index, Economy. Private AI investment by geography, 2025, U.S. dollars. U.S. value: $285.88 billion, rounded to $285.9 billion on this page. Private investment is distinct from revenue, GDP contribution, and total corporate investment.
- Brynjolfsson, Li & Raymond — Generative AI at Work. NBER Working Paper 31161, April 2023, revised November 2023. Study of 5,179 customer-support agents; average productivity measured as issues resolved per hour. Findings are specific to the studied setting.
Sources checked November 3, 2026. Figures are a dated snapshot, not a live data feed. Practical examples and guidance are Skar’s interpretation.