A lot of small and medium-sized businesses are making AI harder than it needs to be.
They hear that they need an “AI strategy”. They see large companies talking about transformation programmes, internal AI policies, model selection, data platforms and multi-year roadmaps. Then they quite reasonably think: “That sounds expensive, risky and probably not for us yet.”
For most SMEs in Gloucestershire and Cheltenham, that is the wrong starting point.
You probably do not need a grand AI strategy before you do anything useful. You need one well-chosen project that solves a real problem in your business.
Not a novelty. Not a chatbot for the sake of having a chatbot. Not an experiment that impresses people for five minutes and then gets forgotten.
A useful AI project starts with a painful, repetitive task that already costs you time, attention or money.
Start with friction, not technology
The most common mistake I see is starting with the tool.
Someone signs up for ChatGPT, Claude, Microsoft Copilot or another AI platform and asks: “What can we use this for?” That is understandable, but it often leads to shallow experiments. A few people try it for writing emails. Someone asks it to summarise a document. Then the business decides AI is “interesting” but not essential.
A better question is:
“What work keeps getting done manually even though it follows a fairly repeatable pattern?”
That might include:
– sorting incoming enquiries
– summarising long email threads before a client meeting
– turning meeting notes into actions
– checking documents for missing information
– drafting first responses to common customer questions
– researching prospects before a sales call
– pulling useful information out of PDFs, spreadsheets or CRM notes
– preparing a weekly internal update from scattered sources
These are not glamorous examples. That is the point. The first useful AI project is often hidden inside ordinary admin, sales, operations or customer service work.
If a task happens often, takes longer than it should, and has a clear “good enough” output, it may be a good candidate.
A useful first project is small, but not trivial
Small does not mean pointless. It means contained.
A good first AI project should have clear boundaries. You should be able to describe when it starts, what information it uses, what output it produces, and who checks it.
For example:
“When a new enquiry comes in, AI reads the message, identifies the type of enquiry, drafts a short internal summary, suggests the next step, and prepares a reply for a human to review.”
That is much better than:
“We want to use AI in customer service.”
The first version does not need to connect to every system in the business. It does not need to make final decisions. It does not need to replace anyone. It simply reduces the amount of repetitive thinking and typing needed before a person can do the important part.
That is where AI tends to work best at the start: as a capable assistant, not an unsupervised replacement.
What this looks like in practice
One of my own experiments was an AI-first CRM concept. The idea was not simply to bolt AI onto a normal CRM as a side feature. I wanted to explore what would happen if AI was built into the working day: helping with tasks, splitting larger jobs into smaller steps, supporting different roles, and assisting with email so the important messages stood out.
That experiment showed something important for smaller businesses. The valuable part was not “AI” in the abstract. The value came from choosing specific jobs where software and agents could reduce manual effort: task handling, email triage, workflow support and role-specific assistance.
The same principle applies on a smaller scale. You do not need to start by rebuilding your CRM. You might start by making one part of your current process easier.
A local accountancy firm might begin with document intake: checking whether a client has sent the right files and producing a short summary for the team.
A professional services firm might begin with meeting follow-up: converting notes into actions, owners and draft client emails.
A trades or property business might begin with enquiries: extracting the job location, urgency, contact details and likely next step from each incoming message.
A local publisher or marketing team might begin with research and drafting: collecting source material, preparing a summary, and creating a first draft for human review.
None of these require the business to “become an AI company”. They require a sensible workflow, clear guardrails and a willingness to test whether the result saves real time.
Keep the human in the loop
For most SMEs, the safest first step is not full automation. It is assisted work.
That means AI can draft, summarise, classify, extract and suggest, but a person still reviews anything that matters before it goes to a client, supplier or member of the public.
This matters because AI can be confidently wrong. It can misunderstand context. It can produce a polished answer that still misses the point. It can also expose sensitive information if tools are used without thinking about privacy and access.
So the first project should include basic rules:
– what information the AI is allowed to use
– what it must not see
– who checks the output
– what happens when the AI is uncertain
– how mistakes are spotted and corrected
– how success will be measured
This does not need to become a 40-page policy before anyone starts. But it does need to be clear enough that staff know how the workflow should be used.
Measure something real
If the first project is useful, you should be able to measure it.
That does not always mean a complicated dashboard. It might be as simple as:
– time saved per enquiry
– fewer missed follow-ups
– faster preparation for client meetings
– fewer documents returned because information is missing
– reduced admin load on a senior member of staff
– better consistency in first drafts or summaries
Pick one or two measures before you start. Otherwise the project becomes another vague technology experiment.
The aim is not to prove that AI is magical. The aim is to find out whether a specific workflow improves the business.
The right question for your business
If you are wondering where to start with AI, do not begin by asking for an AI strategy.
Ask this instead:
“What is one repetitive task in our business that a capable assistant could help prepare, summarise, check or draft?”
Then choose one version of that task. Make it narrow. Add human review. Test it with real examples. Measure whether it helps.
If it works, you will have something much more useful than a strategy document. You will have evidence. You will understand where AI fits in your business, what risks need managing, and what kind of implementation is worth doing next.
That is how sensible AI adoption starts: one useful project at a time.
If you run an SME or professional business in Gloucestershire or Cheltenham and want help identifying your first practical AI project, I can help you map the opportunities, choose a sensible starting point, and build a workflow with the right guardrails. Book a short AI consultancy call or get in touch and ask: “What could AI usefully do in our business?”
