Most small businesses already use AI. Far fewer use it well.
A recent Entrepreneur article by Grant Freeman, president of Thryv, puts numbers on the gap. Adoption among small businesses climbed from 55% to 66% in a year. 70% report more revenue, and 92% say AI saves them time.
Yet 70% also say they need more training.
Comfort has outrun competence. That gap is where the savings leak out, so it's worth a closer look.
Comfortable is not the same as competent
According to the article, 86% of small business owners feel comfortable using AI. Comfort is easy. Knowing what you're doing is the hard part.
It shows in how people learn. The most common sources are YouTube and social media (57%), online resources (49%), and the tools themselves (31%). That's a lot of piecemeal learning.
Piecemeal learning has a predictable failure mode. You automate something quickly, it runs smoothly, and it quietly sends the wrong information to the wrong people. Automation doesn't fix a mistake. It repeats it faster.
Four habits that close the gap
The article suggests four ways to get smarter about AI. Here's how I'd apply each one if your goal is to cut costs in how your business handles information.
1. Start with a business problem, not a prompt
"What can AI do?" is a hobby question. "Where do we lose hours every week?" is a business question.
The best candidates are jobs where information moves from one place to another by hand:
- Copying details from invoices or emails into a spreadsheet
- Turning call or meeting notes into follow-ups
- Sorting a shared inbox
- Writing the same five replies over and over
Pick one. Write down roughly how many hours it eats each week. That number is your baseline, and it's what makes any savings real instead of a feeling.
Watch: Problem-Framed Prompt vs. Vague Prompt (2 min)
See the difference between a vague question and a business problem in action.
[Video embed: demo-habit-1-problem-framed-prompt]
Ready to move from comfortable to competent? → Start the course
2. Put learning on the calendar
If learning only happens when things are quiet, it never happens.
Block 30 minutes a week. Use it to try one new thing on one real task, then note what worked. Small and regular beats a heroic weekend binge.
Watch: Calendar + Email for Weekly Learning (2 min)
See how one ops manager blocks 30 minutes every Monday—and stays consistent.
[Video embed: demo-habit-2-calendar-learning]
Want the templates? → Start the course
3. Trust, but verify
Treat AI like a smart intern, not an expert. It's fast, eager and often right. It can also be confidently wrong.
The question to ask is how cheap it is to check. Reviewing a draft summary takes seconds. Reviewing a batch of customer refunds does not. Build a quick human review step into anything where a mistake costs money or trust.
Watch: When AI Gets It Wrong & How You Catch It (2 min)
A real mistake, a review step that catches it, and why this matters.
[Video embed: demo-habit-3-trust-but-verify]
Build review into every workflow. → Start the course
4. Learn from peers in your industry
Other people in your field have already hit your problems. Ask what they automated, what broke, and what they stopped doing. An hour of that is worth ten generic tutorials.
Where to get real AI education
This is my favorite part of the article. Instead of leaving you to piece things together, it points to structured, mostly free places to learn:
- OpenAI Academy's small business learning track
- Anthropic's AI Fluency Framework and Foundations
- The U.S. Chamber of Commerce's free practical AI education program
- Local community colleges, many of which are expanding their AI offerings
The line I keep coming back to is this one: small businesses don't need to become AI engineers. They need to become AI-literate.
That's a much smaller mountain to climb. You don't need to understand how a model works. You need to know what to hand it, what to check, and what never to give it.
My advice is to pick one structured source and finish it, rather than sampling ten videos. Depth beats variety here.
A simple first month
If you want a concrete way to start, try this:
- Week 1: Choose one repetitive information task and measure how long it takes now.
- Week 2: Try AI on it with real examples. Leave out anything confidential until you know how the tool handles data.
- Week 3: Add a review step. Decide who checks the output and what they look for.
- Week 4: Compare the new time to your baseline. Keep it, change it or drop it.
One task, one month, one number. If it saves time, you've found your first real cost cut, and you know how to find the next one.
The takeaway
AI doesn't cut costs by itself. Knowing what to point it at, and how to check its work, does.
If you'd like a guided path through that, it's exactly what the course is built for: hands-on lessons on putting Claude to work on real operations, with no coding required.
Source: Entrepreneur, "Most Small Businesses Are Using AI. The Real Challenge Is Learning How to Use It Well", by Grant Freeman. Statistics are as reported there.