I run a small solutions practice in Charlotte. Roughly half of my last twelve months has been building AI into things — for my own clients, and as a subcontractor for bigger shops that suddenly need "an AI strategy" by Friday. So I've seen a lot of what works and a lot of what doesn't, on real budgets, with real owners watching.
Here's the honest version, because I'm tired of the LinkedIn version.
Where AI is quietly making small businesses real money
These are the categories where I've personally watched the math work out — usually within a quarter.
Customer-facing copy and creative
This was the first thing to land for small business, and it's still the biggest. Product descriptions for an e-commerce catalog of 800 SKUs. Service-page rewrites. Local landing pages for every neighborhood you serve. Email subject lines you A/B-test against your own old ones. Social captions for a small team that hates writing them.
The trick is that AI isn't replacing your voice — it's giving you a first draft so you can edit instead of stare at a blank screen. A good editor with a halfway decent prompt does the work of three contract copywriters at maybe 5% of the cost. Margins matter for small businesses, and this one is real.
Inbox and intake triage
Tiny businesses get drowned by their own inbox. Bookings, refund requests, billing questions, sales inquiries, vendor pitches, spam, the occasional crisis — all in the same Gmail.
An AI layer that reads inbound mail, classifies it, drafts a response, and surfaces the urgent stuff is genuinely transformative. Not "answer everything for you" autonomous — that's still a year or two away for most workflows — but "make sure your owner sees the three things that matter at 9 a.m. instead of buried under fifty newsletters." That alone is worth a couple of grand a month in saved owner time.
Internal search
If your business has a Dropbox, a Google Drive, a Slack history, and a stack of PDFs nobody can find, an internal retrieval system over all of it is one of the highest-leverage projects you can do. Most small businesses don't realize how much they already know — they just can't find it.
Cost is modest. Tools are mature. Payoff is daily.
Voice and transcription
I'm shocked how many small businesses still pay for human transcription, take meeting notes by hand, or — worse — don't take notes at all. Real-time meeting transcripts with summaries and action items are essentially solved. If you do consulting, coaching, real estate, sales calls, or anything where what was said matters next week, this is a no-brainer.
Image and short-form video generation
Two years ago I'd have told a small business "don't bother." Today, for product photography, social posts, ad creative, mockups, and short b-roll, the tools are good enough that a hands-on owner can produce in an afternoon what used to take a vendor a month.
Where it keeps burning money
Now the unfun part.
"An AI chatbot for our website"
I have built these. I have watched them ship. I have watched owners' enthusiasm collapse around month two when the chat logs reveal that 80% of the conversations are visitors trying to figure out if they're talking to a human, the bot confidently giving wrong answers about hours and pricing, and exactly zero booked leads.
A general-purpose chat widget on a small-business homepage is almost always a worse experience than a clear phone number and a simple form. The exceptions are very narrow: e-commerce stores with deep product catalogs, businesses with extensive self-service documentation, or specific workflows like booking. If you don't have one of those, skip it.
"Replace our customer service rep"
If you have one customer service rep, AI is going to make them faster and let them handle more volume. It's not going to let you fire them. Anyone who tells you otherwise has never run a small business through a real customer complaint.
"An AI agent that runs our marketing autonomously"
This will be true eventually. It is not true today. What's true today is that an AI-assisted marketer with the right tooling can do roughly three times the volume of an unassisted one. That's a huge win — but you still need the human.
"Custom-trained models on our data"
For 95% of small businesses, you don't need fine-tuning, you don't need a custom model, and you definitely don't need to run anything on-prem. You need a good prompt, a good retrieval layer over your existing documents, and someone who knows which API call to make. Anyone quoting you six figures to "train a model on your business" is either misunderstanding what you need or hoping you do.
How to think about it
The pattern I've watched succeed, over and over: pick the most boring, repetitive, owner-time-consuming part of your business. Build AI assistance for that one thing. Measure the time it saves. Then do another one.
The pattern I've watched fail: pick the most exciting, futuristic, AI-on-the-LinkedIn-feed thing. Spend six months building it. Discover nobody asked for it.
Small business AI in 2026 isn't science fiction. It's a calculator that writes English. Treat it like a calculator — useful for specific math, useless for telling you what business to run.
Got a project in mind?
If something here sounds like the work you need done, drop me a line.
Get in Touch