One Employee Beats a Bigger Budget
The biggest disadvantage in AI isn’t having less money. It’s being the only employee.
The data on this surprised me, and it contradicts the thing almost everyone assumes.
The JPMorganChase Institute did something unusual. Instead of asking small business owners whether they use AI, it tracked whether they paid for it — following transactions across 4.6 million small businesses from 2019 through the end of last year.
Here’s what stood out.
A business with one employee making under $250,000 a year adopted AI more often than a solo business making over $250,000.
Less revenue. More adoption. The only difference was a second person.
Across the whole sample, firms with employees adopted at 26.1% against 15.3% for firms without. And the overall number is lower than any survey will tell you: 17.7% of small businesses have ever paid for an AI service, a figure the Census Bureau independently puts at 17.8%. Surveys report two and three times that, because saying you’ve used ChatGPT and paying for it out of the business account are different things.
Why It Matters
If money were the obstacle, the higher-revenue solo firms would be ahead. They’re behind. What the employer firms have isn’t a bigger budget — it’s somebody who can spend Tuesday afternoon setting a system up while the business keeps running.
The solo owner has no Tuesday afternoon. They answer the email, take the calls, deliver the work, chase the invoice, and somewhere in the leftovers they’re meant to redesign the operation around a technology that changes monthly.
Every owner can afford the software. Almost none of them have an afternoon free to implement it.
Which points at something worth naming, because once you can name it you can manage it.
The Almanac Lexicon
Implementation Surface
The total number of separate systems, steps, and connections a business has to build and then keep alive. Every tool you add expands it. Solo owners carry the most of it and have the least capacity to service it — which is why they stall, and why buying more software makes it worse.
Security people have talked about attack surface for years: every additional door is another thing to defend. Implementation surface works the same way. Five tools doing five jobs isn’t five times the capability. It’s five logins, five sets of settings, five things that break in a week you didn’t plan for.
The spending data shows who’s winning that trade. Market-wide monthly spend fell from about $80 in 2022 to $30 in 2025, which reads like AI got cheaper. It didn’t. Firms that adopted in 2019 grew their spending to roughly $90 a month. What changed is who’s in the pool: 63% of business AI users now sit under $40 a month, and only 30.8% of them pay in a given month at all. A small group keeps going deeper. A much larger group is holding one cheap subscription they rarely open.
Tomorrow’s forecast, in your inbox.
One email. Five minutes. Written for owners, not engineers.
Who Should Pay Attention
If you run your business alone, this is about you directly. The distance between you and the two-person shop down the road is widening for reasons that have nothing to do with how good you are at the work itself.
Industry matters too. Professional services sits at 30.3% adoption and information at 39.3%, while construction is at 8.9% and transportation at 5.4%. Lower group means further behind the frontier — and further ahead of your actual competitors the moment you move.
The Owner’s Forecast
Eighteen months from now, the businesses that pulled ahead won’t be the ones with better AI. They’ll be the ones that finished twelve small automations while their competitors researched fifty tools and shipped none.
Same industry. Same budget. Same intelligence. One removed a recurring task from the owner’s week every month for a year; the other kept a browser tab open on comparison articles. By next winter those two businesses will not resemble each other, and the owner of the second one will still believe the problem was that they picked the wrong platform.
The report’s authors land nearby: as access becomes universal and costs stay low, they write, what separates firms depends less on whether they adopt AI and more on how well they integrate it. Adoption is becoming table stakes fast — new businesses started in 2025 reached 10% adoption within six months, a mark that took the 2019 cohort more than six years.
Here’s the uncomfortable part. If bandwidth is the constraint and you’re solo, you cannot fix this by finding more time. There isn’t any. The only move available is to shrink the implementation surface — fewer systems, fewer steps, fewer things to keep alive.
The owners who come out ahead won’t be the ones who tried the most tools. They’ll be the ones who finished one.
Your Next Move
Pick a single task you repeat every week. The follow-up after a consultation. The invoice reminder. The intake form you retype into three places. One task, not five.
Good. Write its steps down in order, on one page. Thirty minutes. Most owners have never done this, which is precisely why the automating never starts.
Better. Automate the part that repeats identically every time — usually the message that goes out after something happens.
Best. Connect the sequence end to end so it runs without you opening anything, then leave the other four tasks alone until this one has run untouched for two weeks.
Whatever you do, don’t spend this week comparing platforms. Comparing platforms is what a busy owner does instead of finishing something.
Today’s Instrument
GoHighLevel
Reduce the number of systems you maintain
Today’s constraint is implementation surface, so the useful tool is the one that shrinks it. GoHighLevel puts triggers, pipelines, and follow-up sequences in one place, which turns the workflow you mapped above into a single thing to maintain instead of four tools wired together and hoped over.
- Best for
- Owners whose bottleneck is follow-up and scheduling, and anyone running separate accounts for clients — sub-accounts are built into how it works.
- Skip if
- You want a polished storefront more than an automation engine. The interface shows its agency origins and the first week asks something of you. If the front end matters more than the plumbing, our Kartra verdict covers the other side of that trade.
Almanac note: this link may earn us a commission at no additional cost to you — it never changes our verdict, and the recommendation always comes first.
Tomorrow: the first automation almost every service business builds — and why it’s usually the wrong one.
Owners Are Asking
Do only 17.7% of small businesses really use AI?
That counts businesses that have paid for an AI service through the business account. Surveys report far higher because they count anyone who says they’ve used it, including a single free trial. Both measure something real; the payment figure measures commitment.
I’m a solo owner. Am I structurally stuck?
No, but the standard advice is wrong for you. Adding tools expands your implementation surface, which is the thing you have the least capacity to carry. Shrinking it is the move that fits the constraint you actually have.
Should I hire someone to implement AI for me?
Finish one workflow yourself first. You’ll learn what you actually need in a way no scoping call will surface, and you’ll be a far better client for it.
Sources: JPMorganChase Institute, “Understanding the use of AI among small businesses,” Wheat, Mac & Passalacqua, April 2026 · U.S. Census Bureau Business Trends and Outlook Survey, 2025