HERNDON, Va. — One of the more persistent frustrations for small business owners over the years has been the cost of custom software. You know exactly what you need. You just can’t afford to have someone build it.
That may be changing.
In Part 1 of this series, SCORE mentors Charlie Morris and Matthew Krieger, presenting at a recent SCORE National Small Business Week webinar, showed how AI can pull insights from the messy data a business already has. In Part 2, they demonstrated AI as a thought partner and as an agent that can execute multi-step tasks autonomously. In this final installment, they take up two more questions: what happens when you ask AI to build something from scratch, and where does AI actually belong in a business?
Building a Tool in Minutes
Krieger, who runs a manufacturing business in Connecticut, wanted to show what he calls “vibe coding” — a term for using AI to write working software through plain-language prompts, no programming knowledge required.
His prompt was simple: build a web app for a services business to track time spent on individual jobs throughout the day. Enter a job name, click “Start” and then click “Stop” when it’s done. Show completed jobs in reverse chronological order. Show a running daily total at the bottom. Make it look modern.
Within a few minutes, Claude Code — Anthropic’s programming tool — had produced a working application. Krieger demonstrated it live, entering a job, starting and stopping the timer and watching the total accumulate. He then asked the AI to add a button to clear the day’s record. It added one. He asked for the daily total text to be larger and bolder. It was done.
“We went from absolute zero to something that worked incredibly quickly,” Krieger says.
He was careful to add the caveats. Software built this way can work fine as a prototype or an internal tool, but it isn’t automatically production-ready. It needs review, it may not be secure and it may not hold up under real-world conditions.
“I don’t know how to build a house,” Krieger says, “but I certainly could build a structure with four walls and a roof. The first time the wind blows, the house is going to go down.”
Still, for a cleaner who needs a simple scheduling aid, an intake tracker, a customer follow-up tool or a job log, the ability to describe what you need and get a working version in minutes represents something genuinely new.
“The bar to experimentation has dropped to near zero,” Krieger says.
Where It Fits, and Where It Doesn’t
The last section of the webinar was the one Krieger called the most important: not whether AI works, but where it should be applied.
He laid out the conditions under which AI tends to deliver real value. Repetitive work with high friction is a good candidate. So is anything that requires synthesizing data from multiple sources. Messy, unstructured data — exactly the kind most small businesses have — is where AI performs especially well. And if AI can enable something you genuinely couldn’t do before, that’s worth serious attention.
The flip side of that list matters just as much. AI doesn’t belong where human judgment and accountability are essential. It’s a poor fit when the process itself is unclear. Implementing AI on top of a workflow you can’t fully explain to yourself is a recipe for wasted time. High-touch customer interaction is another poor fit. And when a simpler tool can do the job, use the simpler tool.
“Don’t implement AI just because it’s there,” Krieger says. “Do it where it fits, not when other technology can do the job.”
Morris added a practical starting point for any dry cleaner who wants to give it a real try.
“Focus on what you want done,” he says. “Pick a task, pick a workflow in your business that you understand well, that would benefit from AI, and go make it happen.”
The data is already sitting there. The thought partner is available around the clock. The tools have gotten easier to use. For operators who have been watching from the sidelines, Krieger’s parting thought was direct: “It’s not cost. It’s not employee count that’s the limiting factor. It’s really just your desire to experiment.”
For Part 1 of this series, click HERE. For Part 2, click HERE.
Have a question or comment? E-mail our editor Dave Davis at [email protected].
