Short answer: no, not so far. In a Federal Reserve Bank of New York analysis published October 8, 2026, small employer firms that use AI were more likely than non-users to expect to hire more people over the next year, and 77% of firms said AI had not changed their labor costs. The honest caveat is that those are expectations from a fall 2025 survey, not measured results.
If you run a plumbing company, a dental practice, a restaurant, or a retail shop in Northeast Indiana, you have probably heard both stories. One says AI will let a three-person shop do the work of ten. The other says it will quietly take away the entry-level jobs your business depends on. You are also hearing from vendors who would like you to buy something.
This post reads the primary research, adds a second survey that tells a more complicated story, is plain about what neither one can prove, and ends with a way to pick a first task for AI that does not put your customers or your staff at risk.
Key Takeaways
- In the Federal Reserve Banks' 2025 Small Business Credit Survey, 46% of small employer firms used AI, and AI users' net expectation of higher employment was 33 points versus 15 for non-users.
- The gap shrank to 10 to 13 points after the New York Fed controlled for firm traits and past hiring. It did not disappear.
- Only 31% of firms reported higher sales from using AI, and 77% saw no change in labor costs. Expectations ran ahead of measured results.
- A separate Bluevine owner survey found 53% say AI already does the work of at least one employee, and 40% are holding hiring flat. That survey is unweighted and vendor-run.
- Pick a first task that is repeatable and has a clear check, keep a person accountable for it, and measure it against your old process.
What Did the New York Fed Find About AI and Small-Business Hiring?
The analysis comes from Will Aarons and Asani Sarkar of the Federal Reserve Bank of New York's Research and Statistics Group, posted on the bank's Liberty Street Economics blog on October 8, 2026. It uses the AI module of the 2025 Small Business Credit Survey, which the Federal Reserve Banks fielded between September and November of 2025. The AI module covered 5,248 employer firms, weighted to match the national population of employer firms.
Here is the adoption picture from the underlying Federal Reserve Banks report, as summarized in the post: 46% of firms with at least one employee said their business or employees were using AI tools, and another 15% planned to adopt them in the next twelve months. Among the users, 63% said AI was somewhat or very important to production, and 51% had integrated it partially or fully into their business processes.
The Numbers From the New York Fed Post
- Employment expectations: AI users had a 33-point net expectation of higher employment over the next twelve months. Non-users had 15.
- Revenue expectations: AI users had a 48-point net expectation of higher revenue. Non-users had 21.
- Labor costs: 77% of respondents reported no change in their firm's labor costs from using AI.
- Sales results: only 31% reported increased sales from their firm's use of AI.
- After controls: the employment gap was 13 points after accounting for firm and owner characteristics and location, and 10 points after also accounting for hiring in the prior twelve months.
- Revenue after controls: a 14-point advantage after accounting for characteristics and past-twelve-month revenue changes.
“Net expectation” is a diffusion index: the share of firms expecting an increase minus the share expecting a decrease. A 33 does not mean 33% of firms plan to hire. It means that among AI users, the people expecting more employees outnumbered the people expecting fewer by 33 percentage points.
One detail matters for owners whose businesses are not thriving. The authors tested whether the result simply reflected cheerful owners using AI. They split the sample by current financial condition. Among owners reporting fair or poor conditions, the employment-expectation gap between AI users and non-users was 21 points, versus 10 points among owners in good condition. For revenue the figures were 24 points versus 15. If general optimism were the whole story, the gap should have been larger among the comfortable firms, not the struggling ones.
What Does This Survey Prove, and What Does It Not?
It proves less than a headline would suggest, and the authors say so themselves. Read the limits before you plan around the numbers.
Five Limits Worth Keeping in Mind
- It measures expectations, not hiring. The New York Fed post closes by saying whether AI-adopting firms actually experience the stronger revenue and employment outcomes they anticipate is a subject for future research.
- It is a snapshot from fall 2025. The survey was fielded between September and November of 2025. AI tools, prices, and habits have moved since then.
- It shows association, not cause. The controls narrow the gap, but they cannot rule out everything else that makes a firm both more likely to use AI and more likely to be growing.
- Expectations are drifting down overall. The Federal Reserve Banks' March 2026 report says expectations for revenue and employment growth both fell to their lowest levels since the 2020 survey: the revenue index dropped from 39 to 33 and the employment index from 26 to 23. AI users are more optimistic than non-users in a more cautious year.
- Results lag expectations. Only 31% of firms reported that AI had increased their sales, and 77% reported no change in labor costs. Plenty of owners are still waiting for the payoff.
The authors offer one possible explanation for the optimism: AI users who said they faced operational challenges with technology, such as their website, social media, e-commerce, or cybersecurity, had a 16-point higher net expectation of revenue growth than AI users without those challenges. The authors read that as a sign AI may help firms get past internal technology limits that would otherwise cap growth, and they conclude it is not simply a story about expanding through hiring.
That last point is easy to miss. The most hopeful group is not the biggest or the best-funded. It is the firms that were stuck on something technical. That fits what we see in conversations with local owners: the first win from AI is often a task they were putting off because nobody on the team had the time or the skill.
Are Owners Using AI to Avoid Hiring Altogether?
A second survey, covered in a Stacker report on Bluevine's Lean Team Economy Report, tells a more mixed story. Bluevine is a small-business banking company. It had Centiment survey 729 U.S. small-business owners between June 25 and June 30, 2026, and it says the data are unweighted. That makes it a snapshot of the owners who answered, not a national measure. It is also a vendor with a product to sell, so we use it for what owners said and nothing more.
What Bluevine's Owners Reported
- 53% said that over the past year AI or automation tools took on work equal to at least one employee's workload.
- 60% are still hiring and developing their existing staff, while 40% have paused hiring or intentionally kept the team small this year.
- 75% plan to hand at least one business function to software or AI instead of hiring for it in the next twelve months. Marketing and sales ranked first, at 39%.
- 65% said their current team size has cost them something, most often time off (33%), followed by their wellbeing (24%).
- 58% named faster decisions as a benefit of a lean team, and 58% named lower overhead and higher margins.
Two details in the report pull against the tidy story. Among owners who had paused or avoided hiring, the top stated reason was keeping operations simple and lean (39%), ahead of uncertain revenue (33%) and high labor costs (30%). Only 19% of those owners pointed to software or AI as their reason, even though most of the full sample said AI already covers at least one employee's workload. Bluevine suggests some owners lean on AI without linking it to a hiring decision, or are hiring for roles AI cannot do, such as sales calls and in-person client work.
The report also says average headcount across more than 160,000 active Bluevine accounts stayed flat over three years while average account balances rose 46%. That is Bluevine's own customer data, so it describes Bluevine's customers and not small businesses in general.
How do you reconcile the two surveys? They ask different questions of different people at different times. The New York Fed looks at firms with employees and asks what they expect over the next year. Bluevine asks owners what they did and plan to do. A reasonable reading is that both can be true: AI users feel more confident about growth, and some are choosing to grow without adding people. What neither survey tells you is what happens to a specific shop on Main Street in Auburn, Kendallville, or Angola.
The strain on owners is worth noticing, too. In the Bluevine data, 65% said running with their current team size had cost them something. A lean team that is also an exhausted team is not obviously a win. Handing off a task to software is only a gain if it gives the owner or a team member time back for work that needs a person.
Where Is AI Actually Doing the Work Right Now?
Adoption numbers vary widely, and the spread says more about methodology than about reality. A TeamShift roundup of small-business AI statistics, updated in October 2026, lists figures from about a quarter to about three-quarters of businesses, depending on the survey. It reports 23.8% of all U.S. businesses using AI in the two weeks ending September 6, 2026 (Census Bureau), 46% of small employer firms in fall 2025 (the Federal Reserve Banks), 66% (the U.S. Chamber of Commerce), and 76% of owners in a Goldman Sachs 10,000 Small Businesses survey. We are citing these second-hand through TeamShift and have not checked each underlying release, so treat the exact figures as the roundup's and use them only to show how wide the range is.
The same roundup passes along a few numbers that matter for anyone who deals with customers. It reports that 7.6% of all U.S. businesses used AI in customer service in the prior six months, according to the Census Bureau, and that 40.0% of businesses planning to use AI expect to use it there. It also reports a Gartner figure from August 2026: 87% of customers say it is essential to be able to reach a human agent when a company uses generative AI for customer service.
That pairing is a useful warning. Owners expect to point AI at the front desk, and customers say they still want a person on the other end. The practical conclusion is not to avoid AI in customer-facing work. It is to use it to speed up the first response and to keep a human easy to reach.
Why Owners Are Looking at This Now
- Hiring is hard. TeamShift reports that 51% of owners had few or no qualified applicants in July 2026 and 36% had job openings they could not fill, citing the NFIB's July 2026 jobs report.
- Time is the scarce resource. Bluevine reports that 48% of owners who use AI save at least four hours a week, from its separate AI Trends Report. That is the vendor's figure, not an independent one.
- Marketing and sales come first. In the Bluevine survey, 39% of owners planning to hand work to software named marketing and sales.
If an unanswered phone call is the thing costing you, that is a problem that exists with or without AI. We covered one version of it in our post on Local Services Ads missed-call fees, where a call that nobody answers can now cost an advertiser money as well as a customer.
What Should a Northeast Indiana Business Do With This?
This section is our judgment, not a research finding. The surveys are national, and none of them isolates Northeast Indiana. What follows is how we would read the evidence if we owned a local service business.
Do not treat AI as a headcount decision. Both surveys point the same direction on one thing: the firms that use AI are not mostly firing anyone. The Census figure TeamShift reports is that 95.7% of businesses using AI said it caused no change in total employment, and Goldman Sachs is reported to have found 87% of owners saying AI augments rather than replaces employees. Treat the tool as a way to give your people back time, and decide on hiring separately, based on your own demand.
Expect the payoff to take a while. Only 31% of firms reported higher sales from AI. If you try it and do not see revenue move in a month, that is normal, not a sign you did it wrong. Measure the time and error rate of the specific task you handed off before you judge the whole thing.
Start where the work is boring and the check is easy. A hypothetical example, clearly not a real client: a small HVAC company gets a few dozen web-form and voicemail inquiries each week, and the office manager spends the first hour of every morning sorting them, replying, and booking the simple ones. That task has a clear definition of done, a person can review the output in minutes, and a mistake is cheap to catch. Compare it with having AI quote a price or promise a technician arrival time, where a mistake becomes a promise to a customer.
Put someone in charge. An AI tool that nobody owns turns into a risk. We wrote about the governance side in our post on shadow AI and approval rules for small firms, and about practical workflow ideas in our guide to marketing automation workflows.
How Do You Pick the First Task to Hand Off?
You do not need a strategy document. You need one task, one owner, and a way to tell whether it worked. Here is a short routine we use when a local business asks where to begin.
A Four-Week First-Task Test
- List five tasks that repeat every week. Think inquiry replies, appointment reminders, review requests, invoice follow-ups, and first drafts of social posts or emails.
- Pick the one with a clear right answer. If a person can tell in thirty seconds whether the output is good, it qualifies. If it needs a judgment call about money or a promise, it does not, yet.
- Write down today's numbers. How long it takes, how many a week, and how often it goes wrong. Without a before, you cannot say whether anything improved.
- Run the AI version beside your normal process for two weeks. A person reviews every output before it goes to a customer.
- Compare, then decide. If the review is catching few mistakes and saving real time, loosen the review on the low-risk cases. If not, change the task or stop.
- Decide where the freed time goes. The Bluevine data suggests owners feel the cost of running lean. Name what the saved hours are for, or they disappear.
It is worth being cautious about anyone who promises a particular number of hours saved or dollars earned before looking at your business. The research in this post is national, mixed, and mostly about expectations. The only number that matters for your shop is the one you measure yourself.
Not Sure Which Task to Hand to AI First?
Button Block helps Northeast Indiana businesses pick one repeatable task, set up the AI version with a person reviewing it, and measure the result against the old process. Tell us what eats your week and we will help you choose.
Frequently Asked Questions
- Does using AI mean a small business will hire fewer people?
- The latest large survey does not show that. In the Federal Reserve Banks' 2025 Small Business Credit Survey, 77% of firms said AI had not changed their labor costs, and firms using AI were more likely than non-users to expect higher employment over the next twelve months. Those are expectations, not results, so treat them as a signal and not a promise.
- How many small businesses use AI?
- It depends on who is asking and how. The Federal Reserve Banks found 46% of small employer firms using AI tools in fall 2025, with another 15% planning to start within a year. A summary of Census Bureau data put AI use at 23.8% of all U.S. businesses in the two weeks ending September 6, 2026, but that figure covers every business, including firms with no employees. The samples differ, so do not compare the numbers directly.
- Did the New York Fed prove that AI causes growth?
- No. The authors found that AI users were more optimistic about hiring and revenue even after accounting for firm and owner characteristics, and they tested whether general optimism explained the gap. But they say outright that whether AI-adopting firms actually experience the stronger outcomes they expect is a subject for future research. The survey measures expectations in fall 2025.
- What is the Bluevine survey, and how much weight should it get?
- Bluevine, a small-business banking company, commissioned Centiment to survey 729 U.S. small-business owners between June 25 and June 30, 2026. Bluevine says the data are unweighted. It is useful for the owner-reported patterns it describes, but it is a vendor survey of a self-selected online panel and is not a national statistic. We cite it for what owners said, not for what is true of all small firms.
- Which tasks should a small business hand to AI first?
- Start with a repeatable task that a person does the same way every time and that has a clear check: sorting and drafting replies to routine inquiries, scheduling reminders, review-request follow-ups, invoice reminders, or first drafts of marketing copy. Keep a person responsible for anything that commits money, makes a promise to a customer, or changes a record.
- Will customers accept AI answering their questions?
- Many want a person available. A Gartner figure summarized by TeamShift says 87% of customers consider it essential to reach a human agent when a company uses generative AI for customer service. Use AI to speed up the first response, then make a human easy to reach.
- How much does it cost to try this?
- It depends on the task and the tool, and we are not going to invent a price. A sensible first step costs mostly time: pick one task, write down the steps and the check, run it for a few weeks beside your normal process, and compare the results before paying for anything larger.
Sources & Further Reading
- Federal Reserve Bank of New York, Liberty Street Economics, “AI Adoption and Employment Expectations: Evidence from a Survey of Small Business Owners” (Will Aarons and Asani Sarkar): libertystreeteconomics.newyorkfed.org (Oct. 8, 2026): the employment and revenue expectation gaps, the controls, and the authors' caveats
- Federal Reserve Banks, “2026 Report on Employer Firms: Findings from the 2025 Small Business Credit Survey”: fedsmallbusiness.org (Mar. 3, 2026): the fall 2025 survey behind the New York Fed post, including the decline in revenue and employment expectations
- Stacker via KRDO, “53% of SMBs use AI to do the work of at least one employee” (Bluevine Lean Team Economy Report): krdo.com (Oct. 14, 2026): the owner survey figures, methodology, and the lean-team findings
- TeamShift, “Small business AI and automation statistics 2026”: teamshift.io (Oct. 2026): a roundup that passes along Census Bureau, Gartner, Goldman Sachs, and NFIB figures, cited second-hand and labeled as such in this post