ai for small business

AI for small business: what's realistic, by industry

For most small businesses, AI helps in the same unglamorous places: first drafts, repeat questions, paperwork, and follow-ups. It rarely replaces the real work, and it always needs a person checking what it produces. How much it helps depends on your industry and on how much repeat work you have.

Below is our honest read on 14 kinds of small business: where AI fits, where to be careful, and where it's not worth it yet.

Last updated: October 10, 2026

The honest numbers

What the research says about small businesses and AI right now, with sources:

1 in 4
U.S. businesses (23.8%) used AI in any part of the business in the past two weeks. The smallest firms lag: under 20% of firms with four or fewer employees, as of May 2026.
Source: U.S. Census Bureau, Business Trends and Outlook Survey, September 2026
1 in 10
Half of small business workers use AI at work, but only about 1 in 10 have been offered formal training.
Source: U.S. Chamber of Commerce Foundation, June 2026
66%
of employees who use AI at work say they've relied on its output without checking it, and only 34% report employer policy or guidance on generative AI.
Source: KPMG and University of Melbourne, 47-country study, 2025
About $30
a month: the typical spend on AI tools for small businesses that pay for them, as of 2025. Cost is rarely the real barrier.
Source: JPMorgan Chase Institute, April 2026
5.4%
of their work hours saved, on average, by workers who used generative AI that week (about 2.2 hours of a 40-hour week). A third saved an hour or less.
Source: Federal Reserve Bank of St. Louis, November 2024 survey
40%
faster on job-style writing tasks with ChatGPT in a controlled experiment, and graders rated the quality 18% higher.
Source: Noy and Zhang, Science, 2023

Our read: most small businesses haven't started, the ones that have are mostly self-taught, and the gains are real but modest per person. They show up on writing-heavy, repetitive tasks, and they depend on people being trained to use and check the tools. That's exactly the gap we fill.

Is AI a fit for your business? A quick test

Good signs

You or your team write the same kinds of emails, quotes, or posts every week.
Customers ask the same questions over and over.
Your records already live in software: email, a booking app, QuickBooks, a CRM.
Someone on your team is curious and has an hour to learn.

Not a fit yet

Your records live on paper and nobody wants to change that. AI needs something to read first.
You want AI to run things with no person checking. It still makes confident mistakes.
The core work is a judgment call (a diagnosis, legal advice, a structural sign-off). AI can help with the paperwork around it, not the call itself.
Every week is completely different. Little repeat work means little to save.
You want AI to decide who gets hired or which tenant gets approved. California now regulates automated decision tools in hiring, and screening is where discrimination risk lives.

Industry by industry

Our fit ratings are judgment calls, not scores: “strong” means lots of written, repeat work and software already in place; “good” means clear wins in fewer places; “with care” means real wins on admin work under strict rules for client data; “limited” means a few office-side wins and not much more yet.

Running a business in California? The California AI rules to know

AI for trades and home services

Strong fit

Plumbing, HVAC, electrical, landscaping, cleaning: lots of quotes, calls, and follow-ups, and the owner is usually on a job when they happen. A strong fit for the office side; the field work stays human.

Where AI helps

Quotes drafted from job photos and a voice memo, in your template, for you to check and send.
Missed calls answered with an instant text back, so the customer doesn't call the next company.
Job notes turned into invoices and follow-up texts, often with the AI already built into field-service apps like Jobber or Housecall Pro.

Be careful: California generally requires everyone's consent before a phone call is recorded, so an AI phone answerer has to say so up front. AI voices on outbound calls and marketing texts both need the customer's permission first.

AI for real estate agents and brokerages

Strong fit

The job runs on writing: listings, follow-ups, updates, and deadlines, most of it repeated for every deal. Most agents have tried AI, but many see no real difference yet; the gap is using it on the right tasks.

Where AI helps

Listing descriptions, social posts, and email blasts drafted from your walkthrough notes.
A personal follow-up drafted for every new lead, waiting for you each morning.
Every date and contingency in a contract pulled into one checklist, for you to verify.

Be careful: The Fair Housing Act bans ads that show a preference for or against protected groups, including ads AI writes (like describing the ideal buyer), so check every word and every fact. Since 2026, California also requires a disclosure, and the original, when listing photos are altered or virtually staged.

AI for accounting, bookkeeping, and tax prep

Strong fit

Document chasing, client emails, and categorizing transactions are repetitive, and the major accounting software now has AI built in.

Where AI helps

Polite, specific reminder emails for missing documents, drafted in seconds.
Transaction categorizing and reconciliation help from the AI features in QuickBooks or Xero.
Client meeting notes and plain-language summary letters, reviewed before they're sent.

Be careful: Tax preparers have strict federal rules about sharing client tax information. IRS guidance from June 2026 also says tax firms should train staff on AI, vet outside AI tools, and review every AI-written document. Keep client data inside approved tools, never a free chatbot.

AI for marketing, design, and consulting firms

Strong fit

Most of the work is already on a computer, so AI fits straight into research, first drafts, and client reporting.

Where AI helps

First drafts of proposals, briefs, and reports, in your house style.
Research and summaries before client calls, plus notes and follow-ups after.
Monthly client reports assembled from your data, then reviewed by a person.

Be careful: Check your client contracts: some say whether AI may be used on their work, and client files stay in approved tools.

AI for insurance agencies

Strong fit

Renewals, quote comparisons, and client questions are repetitive and heavy on reading.

Where AI helps

Renewal reminders and follow-up emails drafted ahead of time.
Side-by-side summaries of quotes for you to explain to the client.
Call notes turned into a summary and a to-do list.

Be careful: Coverage advice stays with the licensed agent, and California regulators warn against AI that treats customers unfairly. Client financial details belong only in tools approved for them.

AI for property managers

Good fit

Tenant messages and maintenance requests repeat endlessly, and most of them only need a quick, correct answer.

Where AI helps

Saved answers to the top questions (rent, parking, noise, move-out), written once and reused.
Maintenance requests sorted by urgency, with the vendor message drafted for your approval.
Lease and notice summaries as a first pass, checked by you against the document.

Be careful: Don't use AI to screen applicants; that's where Fair Housing risk lives. California tenant rules on notices still apply word for word.

AI for shops and online sellers

Good fit

Strong for online sellers: listings, photos, and customer questions add up fast, and platforms like Shopify and Square have AI built in. More limited for walk-in-only shops.

Where AI helps

Product descriptions written from your notes, in your brand voice.
Product photos cleaned up and backgrounds removed in minutes.
Replies to the same customer questions (shipping, sizing, returns) drafted for you to send.

Be careful: Every product claim still has to be true. AI happily invents features, sizes, and materials, so check before you publish.

AI for restaurants, cafés, and caterers

Good fit

The real work happens in the kitchen, but reviews, posts, catering requests, and the same five questions eat the owner's evenings.

Where AI helps

Review replies that sound like you, drafted in seconds for you to post.
Catering quotes and menus built from a forwarded email thread.
A week of social posts written in one sitting, plus menu descriptions that sound like your food.

Be careful: Never let AI write allergen or nutrition claims without a person checking them against the actual recipe.

AI for salons, barbers, and spas

Good fit

Hands are busy all day, so DMs, no-shows, and posting fall to the evening.

Where AI helps

Reminders and deposit rules switched on in the booking app you already use.
Saved replies for prices, hours, and walk-ins, so DMs take seconds.
Captions for a week of Instagram posts, written in one sitting.

Be careful: Keep client notes like allergies and formulas inside your booking software, not in a free chatbot.

AI for fitness studios and gyms

Good fit

Schedules, member emails, and social posts repeat every week.

Where AI helps

Class schedule posts and newsletters drafted for the whole month.
Win-back emails for members who stopped coming, written for you to send.
Answers to the common questions about classes, pricing, and first visits.

Be careful: Don't let an AI tool give medical, injury, or diet advice to members. Point those questions to a person.

AI for nonprofits

Good fit

Small teams carry grant writing, donor thank-yous, newsletters, and volunteer scheduling on top of the mission.

Where AI helps

First drafts of grant answers, built from your past applications.
Personal donor thank-you notes and newsletters, drafted in your voice.
Volunteer schedules and reminder emails sorted out faster.

Be careful: Donor information is private. Some funders also ask whether AI helped write an application, so check their rules.

AI for small law offices

Good fit, with care

Real time savings on intake, summaries, and routine letters, under strict duties of confidentiality and competence.

Where AI helps

Intake calls and forms summarized into a clean first-look memo.
First drafts of routine letters, in your format, reviewed by an attorney.
Long documents summarized as a starting point for your own reading.

Be careful: Lawyers have been sanctioned for filing AI-invented case citations. Bar guidance calls for understanding the tool, protecting client confidences, and checking every word.

AI for dental, medical, therapy, and wellness practices

Good fit, with care

Big wins at the front desk (reminders, questions, marketing), strict rules for anything touching patient information.

Where AI helps

Appointment reminders and confirmations through your practice software.
Answers to the common front-desk questions: hours, insurance, first visits.
Newsletters and social posts that never include patient information.

Be careful: If your practice is covered by HIPAA, patient information only goes into AI tools whose vendor has signed a business associate agreement with you. Clinical decisions stay with clinicians, and California requires a notice when AI writes clinical messages to patients without a provider's review.

AI for contractors and construction

Limited for now

Construction is one of the industries least likely to use AI today. The wins are real but office-side only: bids, change orders, daily logs, and client updates.

Where AI helps

Bids and change orders drafted from your notes, in your format, for you to price-check.
Daily notes turned into client updates and subcontractor schedules.
Job photos organized into a clear report for the client or an insurer.

Be careful: AI can draft paperwork, never sign off on anything structural, code-related, or safety-related. California home-improvement contracts also have required terms, so a person checks every draft.

What it looks like: four made-up examples

These are invented composites, not clients, so you can picture the before and after. No names, no made-up numbers.

Made-up example

A six-person HVAC company in the South Bay

Before: The owner writes quotes at the kitchen table after dinner, and calls go to voicemail while the crew is on a roof.
What changes: Techs send job photos and a voice memo; AI drafts the quote in the company's template and price list. Missed calls get an automatic text back, and urgent words like “no heat” or “gas smell” go straight to the on-call tech.
Who checks: The owner checks every price before a quote goes out. Diagnosis and pricing stay human.
After: Quotes go out the same day instead of days later, and evenings are free again.

Made-up example

A two-chair salon in Oakland

Before: No-shows leave holes in the week, and DMs about prices arrive at 11pm.
What changes: Deposits and reminders get switched on in the booking app the salon already pays for. Saved replies cover the top ten questions, and a week of captions gets written in one sitting. An AI phone receptionist got skipped: the call volume didn't justify it.
Who checks: The owner still posts and replies; AI only does the first draft.
After: Fewer empty chairs, fewer late-night DMs, and posts that actually go out.

Made-up example

A twelve-person accounting office on the Peninsula

Before: Half the staff quietly uses personal ChatGPT accounts, sometimes with client details pasted in. Nobody agreed on rules.
What changes: One team training on the job each person does, a one-page set of AI rules, business accounts that don't train on their data, and saved templates for chasing missing documents.
Who checks: Preparers review everything, and tax information stays inside approved systems.
After: Everyone knows the rules, the risky habit is gone, and document chasing takes minutes.

Made-up example

A family restaurant in San Francisco

Before: Reviews go unanswered for months, and catering requests sit in the inbox until the weekend.
What changes: AI drafts review replies in the owner's voice, and catering quotes come together from the customer's own email thread.
Who checks: The owner reads every reply and every quote before it's sent.
After: Reviews get answered weekly, and catering customers hear back the same day.

Where to start, whatever your industry

Write down every task you and your team did last week. Circle what repeats, what everyone hates, and what ate the most hours. Pick ONE that's low-risk if it's slightly off, and start there. That's the same method we use on the free call.

Book your free 30-minute intro call How we pick what to automate

ready when you are

Start with a free 30-minute call.

Two minutes of quick questions so we come prepared, then we reach out to set up your free phone call. No pitch, no jargon: we hear how your business runs and point you the right way.

Why is the first call free?

We'd rather prove it than pitch it. The first 30-minute call is free so you can hear what's possible for your business before spending a dollar. After that it's $100 an hour for online sessions, training, and builds, low for the Bay Area on purpose: the AI does the heavy lifting. You'll know the price of anything before we touch it.

Book your free 30-minute intro call

Questions before you book? Text us: +1 650 679 0216

Want to know who you're meeting first? Meet your consultant