Ask a small business owner whether the company uses AI, and the answer is usually "a little." Ask the team the same question, and you often get a different answer.
Forbes published a piece on 9 November on why small businesses are struggling to fully integrate AI. Its headline finding is that AI use is spreading fast while real integration into how the business runs stays rare. The part I would ask every owner to read twice is quieter. In the U.S. Chamber of Commerce Foundation poll it cites, more workers said AI adoption was driven by employees trying tools on their own than by direction from their company. Roughly one in ten had been offered any formal training.
In other words, in many firms AI did not arrive through a decision. It came in through the side door.
What the side door looks like
It is rarely dramatic. A coordinator pastes a customer list into a free chatbot to clean up the formatting. A sales rep asks a tool to draft a proposal and sends it with a figure nobody checked. Someone summarizes a vendor contract and makes a decision on the summary. A bookkeeper tries a feature that was switched on by default in software the company already pays for.
None of these people are doing anything wrong. They are trying to get through the week. But the owner now carries risks nobody chose: customer and financial data in tools nobody vetted, output that goes to customers without a second set of eyes, and no record of what was decided with AI and what was not.
In the same poll, privacy and security worries topped the list of concerns workers named. The people using the tools are uneasy too. They just have no rules to follow.
Banning it is not the answer
The instinct is to send a memo that says "stop." It rarely works, and it throws away what the team has already learned. The people who found these tools on their own have done the first round of experiments for free. They know where the time goes. That is useful knowledge, and it walks out the door if the message is "you did something wrong."
The better move is to bring AI in through the front door: on purpose, with rules, owned by someone.
The uncomfortable part
There is a second finding in the same article that owners should not skip. Census research it describes links broader AI integration to stronger business performance. That is a correlation, not proof. Growing firms may simply have more to invest.
The same research found that AI-related job reductions were rare overall. But where firms deployed AI across more functions and invested more heavily in it, it was tied to reductions in staff. Light use by individual workers showed no such link.
I raise this because it changes how an owner should run the project. If you are going to rebuild a workflow, be honest with the people who run it, early. Decide in advance what happens to the time you free up: more customers served, faster quotes, less overtime, or a role that changes shape. Teams can handle change. What they cannot handle is finding out after the fact.
Five steps to bring AI in through the front door
- Take an inventory, without blame. Ask everyone which AI tools they use, for what, and with which data. Then check which AI features are already switched on in your accounting, CRM and email software. You will learn more in one week than in a month of vendor demos.
- Write a one-page rule. Which customer and financial data may go into which tools. Who checks AI output before it reaches a customer. Who is accountable when it is wrong. One page, in plain language, signed off by the owner.
- Pick one workflow you can measure. Quote turnaround, invoice coding and after-hours customer questions are good candidates. Write down how long the job takes today and how often it goes wrong. Without that baseline, nobody will ever know if the tool helped.
- Name an owner and talk to the team. One person owns the result. The people who run the workflow hear the plan, the reason for it and what it means for their work, before anything changes.
- Check back in 90 days. Did time, errors or revenue actually improve? If yes, move to the next workflow. If not, stop, learn and adjust before you spend more.
The order matters more than the tool. Most of the steps above cost nothing except attention.
A note for advisors
If you are a CPA, attorney, insurance broker or banker, your clients' side-door AI is quietly becoming your concern too: financial data in unvetted tools, contracts summarized by a chatbot, coverage questions nobody has asked yet. A simple question at the next meeting opens the door: "Does your team have a written rule for what data can go into AI tools?" Most owners will say no. That is the start of a useful conversation.
Get it right, not quicker
When Goldman Sachs launched its Small Business AI Institute in October, CEO David Solomon told owners, "The important thing is to get it right, not to get it quicker." For most growing businesses, getting it right starts with admitting AI is already inside the building, and then deciding, on purpose, how it should work.
Where did AI come into your business: through the front door or the side door?
Source: Forbes, "Why Small Businesses Are Struggling To Fully Integrate AI," Rahul Dogra, 9 November 2026 (updated 10 November 2026), and the U.S. Chamber of Commerce Foundation and Census Bureau research it cites.