Short answer: start with two or three frequent, low-risk tasks such as drafting emails and summarizing documents, set simple rules for what data staff may share with AI tools, train the team on their real work, and measure time saved before expanding to automation or custom software.
Where small businesses are today
Most small businesses have tried AI, but few have gone deep. Intuit's 2026 AI Impact Report, drawn from more than 34,000 survey responses, found roughly 7 in 10 businesses use AI regularly while only about 1 in 10 invest in dedicated AI tools. The barriers it names most are data privacy, accuracy, and limited knowledge of what AI can do, with cost lower on the list (Intuit). Australia's National AI Centre likewise found trust was the single biggest reason businesses were not adopting AI (ai.gov.au).
The plan below is built to answer exactly those concerns: privacy, accuracy, and know-how.
Days 1–30: pick the first tasks and set the rules
- List repetitive tasks. Ask each person what they write, copy, or summarize every week.
- Pick two or three. Frequent, low risk, and easy to check, such as email drafts, meeting summaries, and job posts.
- Choose one assistant and buy business accounts, which typically do not train on your data.
- Write a one-page AI use policy: approved tools, data that must never be pasted in (customer financials, health information, passwords), and the rule that a person checks anything before it reaches a customer.
Days 31–60: train on real work
- Run a hands-on session where each person uses AI on their own tasks, not on demo examples.
- Collect the prompts that worked into a shared library.
- Name one person as the go-to for questions.
- Track time saved on the chosen tasks, even roughly.
Days 61–90: automate and look at your software
- Automate one workflow end to end, such as lead intake or a weekly report, with a person approving the output.
- Run a SaaS audit to see which subscriptions AI could replace or consolidate.
- Review results: hours saved, quality, and what staff want next.
- Decide the next quarter's priorities from the numbers, not from vendor pitches.
Common mistakes
- Buying tools before choosing tasks. Start from the work, then pick the tool.
- No rules. Staff will use AI either way; a policy makes it safe.
- Trusting output unchecked. AI can state wrong facts confidently. A person reviews anything customer-facing.
- One-off training. Skills fade without practice and a shared prompt library.
Want help running this plan? Our SaaS audit and roadmap and team training follow this same structure.
Common questions
What is the first thing a small or medium business should use AI for?
Writing and summarizing: drafting emails, customer replies, proposals, and job posts, and summarizing long documents or meeting notes. These tasks are frequent, low risk when a person reviews the output, and save time immediately.
How much does it cost to start using AI in a small or medium business?
Business plans for the major AI assistants cost tens of dollars per user per month. The bigger cost is staff time to learn them, which is why short, hands-on training on real tasks gives the best return.