Getting help with AI as a small business: what an AI consultant actually does

If you run a small business and keep hearing that you should be "doing something with AI", you've probably wondered whether it's worth paying an AI consultant to help. This guide explains what the work involves, when you can manage without one, how the pricing usually works, and what to ask before you hire anyone.

What an AI consultant actually does

Much of the work is finding out where your team's time goes and which tasks are worth changing. The AI is often a small part of it.

In practice, a good consultant starts by sitting with the people who do the work. They watch how invoices come in and get entered, how the monthly figures are put together, how missing paperwork is chased and which customer questions get answered the same way every week. From that, they pick one or two tasks where a change would save real time and wouldn't put anything important at risk.

Then they decide whether a tool you already have can do the job, or whether something needs to be set up or built. Often the answer is a setting in the accounting package or a better-organised inbox, and no new product is needed. Where something does need building, they build it around the way your business works, test it on real examples from your own files, and show your team how to use it and how to check what it produces.

A good consultant also writes down how everything works, where it runs and how to change it, so that you aren't dependent on them once the work is finished.

When you probably don't need one

Plenty of small businesses can get a lot out of AI without paying anyone.

If you mainly want help writing emails, summarising long documents or drafting content, the general AI assistants may be all you need. So may the AI features now being added to the software you already pay for, from your accounting package to Microsoft 365. Before switching those on, it's worth asking the supplier a few questions about their new AI features.

If you'd like to learn the basics first, there's free, publicly funded help:

And if you can't yet name a task that's costing you real time, it's too early to call anyone. Spend a fortnight noting the jobs that take more than half an hour and come round every week or month.

When outside help is worth it

It's usually worth getting help when the task involves more than one system, for example invoices arriving by email, details typed into the accounting package, and a spreadsheet updated at the end of the month. Joining those up is often where time can be saved, and where mistakes tend to creep in.

It's also worth it when the work touches customer, staff or financial data, because it has to be set up carefully and you need to know where that data goes. The same applies if you've tried the tools yourself and the change didn't stick, or if nobody in the business has the time to own it.

One way to decide: if getting it wrong would cost more than getting advice, such as a supplier paid twice or a customer's details sent to the wrong place, it's worth talking to someone.

How pricing usually works, and where it goes wrong

You'll come across a few different ways of charging. Some consultants charge a day rate. Others give a fixed quote for a defined piece of work, or charge a monthly retainer. Some build on a software platform and charge a licence for each person who uses it.

None of these is wrong in itself, but there are some common traps:

  • per-user licences that go up every time you hire someone, whether or not that person ever uses the tool
  • a consultant whose income depends on you staying on their platform, which makes it hard for them to recommend anything else
  • a retainer with no end date and no agreed output
  • software costs folded into the fee, so you can't see what you're paying for

Good pricing is usually simpler: a free first conversation, a written scope and quote before any work starts, software costs shown separately and in your business's name, and a clear point at which the work is finished. This is general guidance, not a comment on any particular firm.

Questions to ask any AI consultant, including us

These are the questions I'd want answered before signing anything. Each has a note on what a good answer sounds like.

1. What would you look at first, and what would you leave alone?
A good consultant names one or two specific tasks and is happy to say what isn't worth changing.

2. Who will own what you build?
That means the code, the settings, the instructions given to the AI, and your data. The answer should be "you", in writing.

3. Could we change it, or hand it to someone else, without you?
Listen for admin access in your name, written instructions and a handover session.

4. What happens if we stop working with you?
There should be written exit terms that say how long notice is, what help you get with the move and how you get your data back.

5. Are you reselling any software, or earning commission from a supplier you might recommend?
Either answer is fine, as long as you get a straight one before you sign.

6. Is anything priced per user, and what happens to the cost as we grow?
You want to know now what the bill looks like with twice as many staff.

7. Where will our data go, and which companies will handle it?
They should name the AI provider and where it processes data, and agree a data processing contract with you. The ICO explains how working with suppliers who handle your data should work.

8. How will we know it's working?
Agree what will be measured, such as hours saved or errors caught, and over what period.

9. What happens when it gets something wrong, and who checks?
A person should check anything that matters before it's acted on, and the system should flag when it isn't sure.

10. What would you advise us not to do?
A good consultant can name something they'd steer you away from.

For what it's worth, the answers we give are that you own what we build, you can change it without us, and you can leave on published terms. Ask anyone you speak to for the same in writing.

Red flags

  • Promises with no task attached. "AI will transform your business" with no named job and no way of measuring it.
  • The same answer to every problem. Every task ends up on one platform, licensed per user.
  • Your data and settings in their account. If everything runs in the consultant's account rather than yours, leaving gets hard.
  • No exit terms in writing. If they won't say how you'd leave, assume it would be difficult.
  • Reluctance to name the AI provider or say where your data is processed.
  • A long, paid "strategy" phase that ends in a slide deck and no working change.
  • Pressure to sign quickly, or to commit to a long contract before trying anything small.

A sensible first step

Illustrative example, not a client: the owner of a small trades business spends most Friday afternoons chasing suppliers for missing invoices and typing the ones that have arrived into the accounts.

Start with one task like that. There are three reasonable ways to begin. If it's simple, try it yourself with the tools you already have. If it crosses systems or touches sensitive data, try a short pilot on one task, with an end date and a decision at the end. And if you don't yet know which task to start with, an assessment of where AI would help can help you decide where to start.

Whichever you choose, before anyone in the business uses AI with customer or staff details, agree a few ground rules. Agreeing a short AI policy for your team is a sensible place to start.

Common questions

What does an AI consultant do?

They find the tasks that take up your team's time, decide whether AI or a simpler change would help, and then set up or build something to do it. A good one also tests it on your own examples, shows your team how to check it, and writes down how it works so you can manage without them.

Do small businesses need an AI consultant?

Often not. If you mainly want help with writing, summarising or drafting, the AI tools you already have may be enough. Outside help is worth it when a task crosses several systems, involves customer or financial data, or nobody in the business has the time to own it.

How much does an AI consultant cost in the UK?

It depends on how they charge: a day rate, a fixed quote for a defined piece of work, a monthly retainer, or a per-user licence for a platform. Whichever it is, ask for a written scope and quote before any work starts, with software costs shown separately and in your business's name.

Is there free AI support for small businesses in the UK?

There's no dedicated free AI service, but free publicly funded business advice is available: the Business Support Service in England, Business Gateway in Scotland and NI Business Info in Northern Ireland. The ICO and the NCSC also publish free guidance for small organisations on data protection and cyber security.

What should I ask before hiring an AI consultant?

Ask who will own what they build, whether you could change it or leave without them, whether they resell software or earn commission, what is priced per user, and where your data will go. Get the answers in writing.

How do I start using AI in my small business?

Pick one task that takes up someone's week and try it with the tools you already have. If it's more involved than that, run a short pilot on that one task with an end date, and agree a few ground rules for using AI with customer or staff data before anyone starts.

Talk to us

If you have one task in mind and want a straight answer on whether AI would help, and whether you need anyone's help with it, book a free 30-minute call. We'll tell you plainly, even if the answer is that you can do it yourself.

This article is general information, not legal, financial or regulatory advice.

Bring us one task that eats up your week

Start with a free 30-minute call. We'll tell you plainly whether we can help and what a sensible first step would be.

Book a free call