Get in Touch
Blog

Small businesses use AI. Few have built it in. Here's how to close the gap.

5 min readWebzita Team

Most small business owners we talk to have tried AI. Someone on the team uses a chatbot to write emails. Someone else summarises meeting notes. A few people pay for their own accounts.

That’s a good start. It isn’t the same as AI being part of how the business runs.

The gap, in numbers

In March, Goldman Sachs 10,000 Small Businesses Voices published a survey of 1,256 small business owners. 76% said they use AI. Only 14% said it’s fully part of their core operations.

The owners also named what’s in the way: not enough technical know-how, trouble choosing tools, and worries about data privacy. 73% said more training and implementation help would make a difference.

A Thryv survey of 561 small and mid-sized business decision-makers, published in July, tells a similar story. 66% use AI, up from 55% a year earlier. 70% say they need more training to use it well, and 57% say their main source of training is YouTube and social media.

These are surveys of the owners who answered them, not of every business in the country, and each one defines “using AI” a little differently. The direction is still clear. Most teams are learning AI on their own, one tab at a time. That’s normal. It’s also where most of the value is still waiting.

Why trying AI doesn’t add up

When AI lives in a chat window off to the side, every task starts from scratch. Someone copies details out of one system, pastes them into a chatbot, then copies the answer back. The time saved on writing gets spent on shuffling.

It also creates risk nobody sees. In IBM’s 2026 Cost of a Data Breach report, shadow AI, meaning tools staff use without approval, was involved in 43% of the breaches studied. That’s more than double the year before. In a PagerDuty survey of 1,250 office workers at large companies, two thirds said they’d used AI tools they believed weren’t allowed, and 34% had put customer data into public AI tools.

Those are big-company numbers. The habit isn’t limited to big companies.

The answer usually isn’t more tools. It’s a few decisions, made on purpose.

Step 1: Pick one job that repeats every day

Start with work your team already does every day, the same way. Not the most exciting idea. The most repeated one.

A dental office might pick appointment reminders and recall messages. A contractor might pick turning site notes into a client update. A property manager might pick sorting repair requests. Different businesses, different first jobs. The approach is the same.

Good first jobs have three things in common:

  • They happen often enough that small savings add up.
  • A person can check the result quickly.
  • A slightly wrong answer is easy to fix.

Write down how the job works today, step by step. You’ll often find a handoff or two worth fixing before any AI is involved.

Step 2: Write a one-page AI policy

Banning AI rarely works. People keep using it and stop telling you. A short, clear policy works better, and it fits on one page:

  1. Use the approved tools, with business accounts. Never personal ones.
  2. Keep private details out of public chatbots. Customer, patient and financial information stays inside.
  3. A person checks anything a customer will read.
  4. Turn on two-factor sign-in for every AI tool, the same as you would for email.
  5. Not sure? Ask before you paste. No one gets in trouble for asking.

That’s it. Share it, then walk through it with the team for half an hour. Adjust it to your own tools and your own data.

Rule 5 matters most. People who feel safe asking make fewer quiet mistakes.

Step 3: Connect only the data the job needs

This is where trying AI turns into AI that’s built in.

A lot of small businesses don’t have a software problem. They have a too-much-software problem. Bookings live in one app, invoices in another, customer notes in a spreadsheet, and the team is the glue between all of it. AI can’t help much with work it can’t see, and copy and paste isn’t a connection.

So connect the systems around the job you picked in step 1, and only those. If the job is sorting repair requests, the AI needs the request, the property and the vendor list. It doesn’t need anyone’s bank details. Start narrow. It’s much easier to widen access later than to pull it back.

Connecting things rarely means ripping everything out. Keep what works, connect it, and retire what’s holding you back. One system people trust beats five that nobody fully believes.

After the first quarter

If these three steps go well, you won’t have an “AI project.” You’ll have one job that runs more smoothly, a team that knows the rules, and systems that share what they need to. Then you pick the next job.

The tools will keep changing. Clear rules, clean data and connected systems carry over to whatever comes next. That’s why it’s worth starting now, and starting small.

A short checklist

  • One daily job, written down step by step
  • A one-page AI policy the whole team has read
  • Business accounts and two-factor sign-in on every AI tool
  • Only the data that job needs, connected
  • A person reviewing anything customers will see

Webzita is a small studio in Austin and Los Angeles. We build custom platforms, websites and apps, and work with businesses on AI integration, automation, networking and security. If you’re working out where AI should start in your business, we’re happy to think it through with you.

Sources

All articles