Most small businesses are already using AI, just not deliberately. Someone uses ChatGPT for emails, someone else tries an AI meeting recorder, a free tool appears in the marketing team, and nobody knows what data went where. Surveys in 2026 suggest around two-thirds of small businesses use AI in some form.

An AI strategy doesn't need consultants or a thick report. You can write a useful one in an afternoon. Here's how.

Hour 1: Goals and pain points

Start with business goals, not AI

Write down your three most important goals for the next 12 months. For example:

  1. Grow revenue 20% without hiring more admin staff.
  2. Respond to every enquiry within an hour.
  3. Cut month-end bookkeeping time in half.

List pain points

Ask the team: what takes too long, what's repetitive, where do mistakes happen, what do customers complain about? Typical answers:

  • Writing quotes and proposals.
  • Following up leads.
  • Answering the same customer questions.
  • Meeting notes and actions.
  • Data entry between systems.
  • Chasing invoices.
  • Creating marketing content.

Hour 2: Use cases and priorities

For each pain point, ask whether AI could help, then score it:

Use caseValue (1–5)Effort (1–5)Risk (1–5)Priority
Instant lead follow-up522High
AI support answers from help docs433Medium
Meeting notes and actions312High
AI-drafted proposals422High
Bookkeeping categorisation323Medium
AI-written blog posts223Low

Pick three to start. High value, low effort, manageable risk. For inspiration, read 12 AI agent use cases small businesses run.

Estimate the payback

For each priority, estimate time saved, errors avoided and revenue gained against costs. Our AI automation ROI guide has a simple method.

Hour 3: Tools, rules and risks

Choose tools

Check what your existing software already includes before buying anything new.

Write the rules (your AI policy)

A one-page policy covering:

  • Approved tools, and that work data only goes into them.
  • What must never be pasted in (customer personal data unless approved, passwords, confidential contracts).
  • Human review for anything customer-facing, legal or financial.
  • Disclosure when customers interact with AI.
  • Who to ask when unsure.

Our AI policy guide has a template.

Check the risks

RiskSafeguard
Wrong answers reaching customersHuman review, grounding in your documents. See AI hallucinations
Data protection breachesBusiness plans, DPAs, data minimisation. See AI and GDPR
Legal transparency dutiesAI disclosure in chat and voice. See EU AI Act Article 50
Over-dependence on one vendorExport options, documented processes
Staff anxietyClear communication: AI removes drudge work, not jobs

Hour 4: Plan, measure and own it

A 90-day plan

WeeksAction
1–2Roll out business AI assistant and policy; 1-hour training session
3–6Launch priority use case 1; measure baseline and results
5–8Launch priority use case 2
7–10Launch priority use case 3
11–12Review results; decide what to scale, fix or stop

Measures

  • Hours saved per week.
  • Response times.
  • Error rates.
  • Revenue or conversion changes.
  • Staff satisfaction.
  • AI tool costs.

Ownership

Name one person responsible for AI: tools, training, policy and reporting. Give them a couple of hours a week.

A one-page template

Our AI goals: [3 business goals AI should support] Priority use cases: [3 use cases, each with expected benefit] Tools: [assistant + targeted tools + automation] Rules: [link to one-page AI policy] Risks and safeguards: [top 5] 90-day plan: [milestones] Measures: [how we'll know it worked] Owner: [name] Review date: [date]

The short version

An AI strategy for a small business fits on one page: start from business goals and pain points, score use cases by value, effort and risk, pick three with clear payback, choose tools (checking what you already have), write a one-page AI policy, plan 90 days, measure results and name an owner.

Want help building and rolling out your AI plan? Our consulting team and AI automation team run practical AI workshops and implementations for small businesses.