What Five Small Business Owners Learned About Where AI Belongs
Michael Quinn spent somewhere between ten and twenty hours working out the profitability of Acorn Products Corporation's customer accounts. Roughly a hundred of them, one at a time, an hour or two here and there.
Then he tried it in Claude. About half an hour, and almost the same result.
"If I had used that to begin with, I would have made so many less mistakes, would have saved so much time just formatting alone," he said. "And it looked better than mine. It was a lot less messy than mine."
Michael is one of five owners we talked to this year for Pulse Check, our conversation series with small business owners. We asked all of them about AI as a new tool that pushes them to ask the age-old entrepreneurship question: what problem does this solve for me? And how do I weigh the cost with the potential opportunity?
Each of them had come to their own answer – and found limits.
Where they pointed AI and where they stopped
Coach Alex Lalaoui: Soccer Learning Center, youth soccer, in what he calls the most diverse place in Jersey City
What he uses it for:
Building the program schedule, and the updates that follow: it "usually takes me forever," and now comes together "within a short period of time"
Communications with families, in more than one language: "believe it or not, also include our multilingual community. That the AI helped me on that as well"
Flyers: "I used to wait for flyer for decades just to get a simple flyer. Why do I have to wait for three days for this person to give me an okay?"
Drafting a message, then finding what he left out: "I was thinking to send message with certain ingredients to it, the one, two, and three. When I put it up in ChatGPT or something, next thing come it gives me four and five and six"
He estimates it now handles more than half his workload, "with eyes closed."
His take: He started last year, twenty-three years into running the business. When he mentioned it to people at US Soccer Federation, one told him it was like he'd been hanging out with George Washington. The method he arrived at hasn't changed since: "I can quickly generate ideas, structures, drafts, then refine them with my own experience and standards." He writes the delicate ones himself first, and the reason he runs them through at all is time pressure: "I know I need to deliver that message quickly, but I'm afraid about just to be put the right word into it."
Where it falls short: AI "does not coach or teach." It can't feel the energy of a session, or sense when a player is struggling at home, or decide whether a child needs encouragement today rather than a correction. That judgment, and the trust underneath it, is "irreplaceable." His summary: "AI handles my inbox and I handle my players."
Terrell Dominick: womenswear e-commerce
What she uses it for:
Copywriting and content
Email responses and automations: the category she reaches for a word for and doesn't like. "Low level, I don't want to use the low level, but I guess I have no other word for it"
Some design work
Closing the gap between an idea and a manufacturer: "helping me bridge the certain things that I'm not aware of. I'm able to close those communication gaps from ideas to manufacturing, which has been amazing"
Her take: Most of it is work she used to pay someone else for. "It's saving me a lot of money there." She got there reluctantly. Her first reaction was to want no part of it, and she described what she saw coming in the voice of whoever would be doing it: "As a black person, a black woman, making us obsolete in certain areas. Like, we don't necessarily need you to do this, or we can steal your creativity, your music, and we can create an AI person to do it." She adopted it anyway. "I don't want to be a dinosaur. At the end of the day, I'm still a business owner."
Where it falls short: She spent seven hours prompting to get one dress onto one AI model. "I joked with my husband, I was like, I could have had seven photo shoots." After all of it, the model suggested she hire a photography editor. "You could have said this seven hours ago. We've been working on the same thing over and over and over." She's open to the possibility that the fault is hers, that there's something more "elevated" she's "missing." What she took from it: "AI is still kind of like fast food. If you're hungry, you can stop and get a really quick hamburger. You won't be hungry anymore. But is it the best hamburger you could possibly eat? I don't think so."
Dr. Sarosh Cooper: Footprint Therapy and Evaluation Services, mental health
What he uses it for:
Progress notes: using AI to transcribe audio and draft progress notes, helping therapists complete documentation more quickly and meet compliance requirements
Scheduling: he built an AI agent to match clients with therapists and let them book available appointments, even in the middle of the night
Billing support to check insurance coverage and deductibles, estimate clients’ out-of-pocket-costs and tell the practice how much to collect
His take: "There's many practical things that AI is taking care of for therapists that before were never done." The scope is narrow on purpose: a tool "not in the sense of therapy, but on the back end." Beyond that he's agnostic. "AI like everything else, it's how you use it."
Where it falls short: The problem he didn't anticipate came from outside. His clients used to look something up before calling; now they turn up having already tried the fix. "It's a whole different world. They've already gone to ChatGPT or to Claude or Grok or whatever and they tried whatever the AI bot has told them. And so there's like this added layer between the client calling us and us being able to help." He isn't against it: "if there's a quick fix for somebody, why not?" But: "The fact is most problems of living, they're not gonna be fixed that way."
Meadow Propsom: Silvia Sol, fashion, manufactured in Colombia
What she uses it for:
Building her Shopify store: using the platform's built-in AI chatbot as something to "ask any questions to"
Breaking an unfamiliar build into steps: "break down the steps that you need to do"
Her take: For her the value was access rather than time. "For a lot of first-time business owners, you really don't know how, you're starting from scratch." Without it: "I would have to probably find someone IRL and ask them, hey, can you help me set up my website." Her rule for where it goes is practical: "it's not going anywhere, so we might as well find the way to optimize it and do the tasks that we don't really want to do." And a limit on how far: "not necessarily overusing it and not relying on it too much either."
Where it falls short: A custom font meant going into the code. "Even with AI, it was still a struggle." The bigger line is at the thing she sells: "AI can't replicate taste." Taste comes from "years of curation or understanding yourself and your style," and AI "can't replicate the eye that you have."
Michael Quinn: Acorn Products Corporation, injection molding, two plants in upstate New York
What he uses it for:
Profitability analysis across roughly a hundred customer accounts
Document formatting across his engineers' output: "a consistent format across multiple documents is what I like it for. We can just have our logo on everything"
Meeting agendas from bullet points he dumps in: "It looks way better, like just way more professional"
Quick reference: "a lot better than Google if you just need to ask a quick question"
Plant layout, while he works out how to get two facilities under one roof
Code he couldn't write, for a submit button in his quoting spreadsheet: "I could have never done it. I would have never had time to do it"
His take: He picked it up at school, where students were taught to use it as a tool and got "berated with more questions" if a paper looked like the machine had written it. "You gotta be careful with what you're doing with it, right? It's not perfect."
Where it falls short: He set up a scheduled task to sweep the web weekly for potential customers. It does exactly what he asked it to, and he stopped checking. "Nothing's came out of it yet," he said. "I haven't even checked it in a few weeks." On the larger question, he's blunt: "when it comes to actually running the business, doesn't do a whole lot."
What’s rising to the surface now
Across these small business owners, a common set of principles around how they leverage AI are emerging.
Start with what's scarce, not what's possible. Terrell knew which one she was short of: "I needed to save me time, not necessarily money. In my business, time is a little more valuable." Her verdict on the tool: "It's saving me money, but I don't necessarily know if it's saving me time." Money saved is worth little if hours are the constraint.
Point it at what you don't want to do. Meadow: "it's not going anywhere, so we might as well find the way to optimize it and do the tasks that we don't really want to do." And a limit on how far: "not necessarily overusing it and not relying on it too much either."
Expect to fail first. Sarosh: "You have to let people make mistakes. So you give the tool, you sit together, you struggle together, and you find a way forward. And you see what comes from it." He started with no goal in mind -- "I kind of go step by step with my staff and see what openings come."
Experiment across tools. Coach Alex uses ChatGPT, Perplexity and Copilot side by side, "always kind of trying to taste a flavor from [a] different one."
Check it against something you already know. Michael had already spent ten to twenty hours on the profitability of a hundred accounts when he ran the same job through Claude. Half an hour later: "I got the same, almost the same result that I came up with." He had a number to compare it against. That's what made the fast version worth anything.
Take work off people, not people off payroll. Sarosh frames it as something an owner has to decide up front: "not let me replace this employee, but let me take that stuff off their plate and see what other talents they have inside of themselves." The staff member whose tasks it absorbed moved into higher-level work.
Ask who pays for the savings. Terrell's version reaches past her own payroll, and she knows it's unresolved: "Are we training our creatives on these systems so we're not just replacing jobs? When that happens, people at the bottom are typically the first ones to go." There's an operational version too -- cutting a role doesn't remove the work. "It's just one person now just working four jobs, the four jobs that you eliminated."
Keep a human at the wheel. Coach Alex: "It is a supplement. It's just like a vehicle that only moves with human at the wheel, guided and shaped by our values and judgments."
The pattern bubbling up across all five: automate what pulls you away from what customers come to you for.
That isn't a rule exclusive to AI. After all, knowing what you're short of, taking an iterative approach, being deliberate about where you take risks, & protecting the people who do the work are foundational to building a business. In the long run, the enduring principles of how to build smart are evergreen, and new technologies will continue to provide opportunities to apply them.
These conversations continue to inform our work around capacity-building for small businesses looking to grow. Explore our offerings here.
