I took four calls this week with business owners about AI. Two of them wanted to know how to get more out of it. Two of them wanted to know how to stop it doing something stupid.
That split isn't a coincidence, mate. It's the entire AI conversation in small business right now, and there's a survey out this month that finally puts numbers on it.
The gap nobody talks about
Bluevine's 2026 SMB AI Trends Report, based on more than 900 small business owners and over 200,000 business banking accounts, published on 15 July 2026, found that 48 percent of small businesses using AI are getting more than four hours a week back. That's half a working day, every single week, gone from someone's plate.
Here's the bit that stops you in your tracks. The same survey found 78 percent of small business owners still don't trust AI to handle basic tasks. Seventy four percent are actively using or testing AI tools. And eighty two percent say they're hitting roadblocks that stop them going any deeper with it.
Read those numbers side by side. Nearly half the SMBs using AI are already clawing back hours every week. Most of them, at the same time, don't trust the thing that's saving them the hours. That's not confusion. That's two different experiences of the same technology happening inside the same business owner.
I see it on calls constantly. Someone tells me their AI tool sorts their inbox brilliantly and then, in the same breath, says they'd never let it touch a client email without checking it first. Both things are true. Both are sort of sensible, and that's the whole point.
Why the trust gap exists (and it's not paranoia)
According to the same Bluevine report, data security and privacy concerns are now the top barrier to deeper AI adoption at 33 percent, up from 23 percent in 2025. Lack of trust in AI's accuracy sits close behind at 31 percent. People aren't avoiding AI because they're behind the times. But they're avoiding deeper use of it because the last twelve months gave them reasons to be careful, and those reasons are getting more common, not less.
A separate 2026 AI and Small Business Adoption Survey from Thryv, published via Business Wire on 9 July 2026, found small business AI adoption has climbed to 66 percent, up from 55 percent a year earlier. At the same time, 70 percent of those owners admit they need more training to use it properly, and a third say they're spending more on AI tools despite a widening skills gap as well.
Put plainly, adoption is racing ahead of confidence. People are buying the tools faster than they're learning to trust or run them well. That's exactly how you end up with an AI reality check moment, the one where a business owner realises they've been paying for something for six months and still couldn't tell you if it's actually working.
The return on investment question nobody wants to answer honestly
So Bluevine's report also found that 52 percent of small businesses using AI report a real return on their investment. The other 24 percent admit they haven't seen any ROI at all and are stuck in what the report calls an inefficient trial and error cycle.
So that split, roughly three quarters getting some result and a quarter getting nothing, tracks with what I hear on calls. The businesses getting a genuine result almost always did one thing differently at the start. They picked a single, narrow, well defined task and measured it. The businesses stuck in the trial and error cycle almost always started broad. They bought a tool because it looked clever, pointed it at "the business" in general, and waited to see what happened.
Nothing happens when you point AI at "the business in general." Nothing ever does. It's not a strategy, it's a hope.
What actually closes the gap
Here is the pattern from the calls this week and from the businesses in these reports that are seeing a proper return.
Pick one task, not one department. Not "AI for marketing." One specific job: draft the first reply to every enquiry, summarise every call, chase every unpaid invoice after day seven. Small and specific beats broad and vague every time.
Decide up front what "working" looks like. Hours saved, replies sent faster, fewer things falling through the cracks. Write the number down before you start. Otherwise, three months later, you're guessing whether it did anything, and guessing is how you end up in that 24 percent with no ROI to show for it.
Keep a human on anything that touches trust. Client emails, pricing, anything with legal or financial weight. The 78 percent who don't trust AI with basic tasks aren't wrong to want a check in place. The fix isn't blind trust, it's a sensible handoff. AI drafts, a person decides.
Review it monthly, properly. Not "is it still running." Is it still doing the job, is the output still accurate, has anything changed that needs a tweak. The businesses getting hours back every week treat their AI setup like a piece of kit that needs the odd service, not a thing you switch on and forget about.
So this is the whole idea behind The Autonomous Business framework I bang on about. Automation on its own isn't the win. Automation with a clear measure of what it is meant to achieve, and a human still steering the direction, is what turns "we're using AI" into "we got four hours back this week and I can prove it."
The actual takeaway
You don't need to trust AI with everything. Nobody serious is asking you to. You need to trust it with one thing, measured properly, with a person still checking the bits that matter. Do that and you're in the 48 percent getting hours back. Skip it and you're gambling in the 24 percent with nothing to show for the subscription.
So four calls this week told me the same thing four different ways. The businesses that are winning with AI right now are not the ones using the most of it. They are the ones who picked one thing, wrote down what success looked like, and actually checked.
Want to see how this could work in your business? Book a call and let's talk about where you're at and what's possible.
Brewed by Steven, poured by Viktor
About Steven Tann: Steven helps business owners build systems that run themselves using AI. After 10+ years helping 7,000+ businesses and building his own autonomous operations, he's the bloke who actually does it, not just talks about it. Find out more at steventann.com.
---
Sources:
- Bluevine, "2026 SMB AI Trends Report" via PR Newswire, 15 July 2026: https://www.prnewswire.com/news-releases/bluevine-study-shows-ai-is-saving-nearly-half-of-smbs-4-hours-a-week-302823855.html
- Thryv, "2026 AI and Small Business Adoption Survey" via Business Wire, syndicated on Wedbush Investor Relations, 9 July 2026: https://investor.wedbush.com/wedbush/article/bizwire-2026-7-9-ai-adoption-continues-to-rise-but-70-say-they-need-more-training-to-use-it-effectively