Hi

If you run a small business and you’re already fairly tech-comfortable, you’ve probably felt this shift.

Two years ago, AI was mostly “interesting”. Now it’s quietly turning into an everyday layer inside the tools you already use: email, meetings, documents, search, bookkeeping, reporting.

And that matters because the advantage isn’t “having AI”. It’s making faster, better decisions with cleaner information.

A Zoom report put some useful shape around this. In 2025, 77% of SMEs said AI has had an impact on their business, but most describe that impact as some benefit rather than a dramatic transformation. That rings true. The firms that win are doing the basics, but better, every week.

Here’s what I think has genuinely changed for the small business community, and what to do next.

1) The “information problem” is being replaced by a “judgement problem”

For years, most small businesses had the same issue: you could find the answer, but it took time. You’d open ten tabs, trawl your inbox, chase someone for “the latest version”, and still be unsure.

AI flips that. It summarises, drafts, and pulls threads together inside your workflow. The report describes this move from searching and clicking around, to getting a summary and first draft where you work.

So the constraint moves.

Less time spent finding and formatting information. More time needed to decide what’s true, what matters, and what to do.

That’s why concerns about bad or misleading information are rising, even as some other worries fall.

This is the new skill: commercial judgement with good controls.

2) The biggest upside isn’t “content”. It’s decision-making

A lot of the public noise is about marketing content. Useful, yes. But the more important shift is that AI is starting to act like a decision support layer.

In the Zoom survey, SMEs ranked improved analysis and decision-making as the top benefit (43% in 2025, up from 39% in 2023). Planning and forecasting also rose (38% in 2025).

That’s where the money is. Because better decisions show up as:

• fewer pricing mistakes
• faster correction when costs drift
• earlier detection of cash squeezes
• quicker learning about what customers actually buy (and why)

Is there a better way to run a business than tightening that feedback loop? Not many.

3) “AI advantage” is becoming a baseline expectation

One uncomfortable truth: once customers get used to faster responses, clearer proposals, and smoother onboarding, they don’t view it as special. They view it as normal.

That’s the real change for small businesses: AI is compressing the gap between small and large in the day-to-day stuff.

Some surveys show very different adoption rates depending on wording and sample. For example:

• The UK ONS has reported around 23% of businesses using some form of AI (late Sept 2025).
• An OECD survey reported generative AI in use in 31% of SMEs, including the UK, in 2024. 

Different definitions, but the direction is clear: this is moving from early adopters to mainstream. So the question becomes: what are your competitors quietly doing with AI that reduces their costs or increases their conversion rate?

4) The “small wins” are the compounding wins

Business owners often overestimate the impact they’ll get from a single AI rollout, then they underestimate what happens when small improvements stack up.

The Zoom report shows this nicely: many businesses report benefits, but relatively few call it a major benefit yet. That’s exactly how change looks early on. In practice, compounding comes from three moves:

1. Remove repeat work
   Summaries, meeting notes, drafting, action lists, follow-ups. The report calls out streamlining repetitive tasks as a key benefit (tied for third at 31%).
2. Reduce “decision latency”
   Time between “problem appears” and “we act”. AI can shorten this if your data is in decent shape.
3. Standardise quality
   Not perfection. Just fewer sloppy emails, fewer missed actions, fewer “we forgot to send the quote”.

5) Costs are still the pressure point, and AI is being treated as a lever

The report notes that controlling or reducing costs sits in the top three business challenges (41% of SMBs). It also says 81% would pay a premium for AI capabilities.

This is a big tell.

Small businesses are effectively saying: “I’ll pay for tools if they reliably buy back time, reduce waste, or improve margins.” That’s sensible. But only if you measure it properly.

A practical way to adopt AI without making a mess

Here’s the operating approach I recommend to owners who want value, not theatre.

Step 1: Pick two workflows, not ten tools

Choose two areas where time is being burned every week. Good examples (generic, but real):

• customer enquiries → quote → follow-up
• supplier invoices → coding → payment run
• management reporting → actions → owners’ decisions

Step 2: Define one “before and after” measure for each workflow

Keep it boring and measurable:

• time to send a quote
• number of invoices overdue
• days to close month-end
• forecast accuracy at 30 days
• gross margin variance flagged within 7 days

Step 3: Put guardrails around truth and confidentiality

Non-negotiables:

• The accounting system is the source of truth, not the chatbot.
• AI outputs that affect money, tax, legal, or HR get human sign-off.
• No client identifiable data goes into tools unless you’re confident about the settings and terms.

(If you want, I can share a one-page “AI data rules” template that’s written for UK small firms.)

Step 4: Build repeatable prompts and checklists

This is where owners miss the trick. Don’t rely on “ask it nicely”. Create standard prompts like:

• “Summarise these meeting notes into actions, owners, dates, and risks.”
• “Turn this draft into a client-ready email with a clear next step.”
• “Explain the movement in gross margin and overheads month-on-month using these numbers.”

Consistency beats cleverness.

Step 5: Review weekly, improve monthly

AI is not a project. It’s an operating habit.

Three questions for you (that decide whether AI helps or distracts)

1. Where is your business currently “slow”?
   Quoting, cash collection, hiring, delivery, month-end reporting, decision-making?
2. What is your source of truth for numbers today?
   Proper cloud bookkeeping, or a mix of spreadsheets and bank screenshots?
3. If we could only improve one metric in the next 90 days, which would genuinely move the dial?
   Gross margin, lead conversion, cash in bank, delivery time, utilisation, overhead control?

If you reply with your answers, I’ll suggest two high-return AI workflows to start with, plus the controls I’d put around them so you get the upside without the downside.

And finally...have a very happy, peaceful and joyful Christmas and thank you for being with me for another year and the feedback you give me. As you know I read every email and truly value all your comments and suggestions.

Noel Guilford