AI consultant for small business: what you actually get

An AI consultant should be able to tell you which parts of your business are worth handing to AI, which aren't, and what it would take to make the useful ones work day to day. For a small business, that last answer counts most, because a strategy document doesn't answer the phone or process an invoice.

Nobody regulates the title, so it covers everyone from a former management consultant with a slide deck to a developer who has spent years connecting software. It pays to know what the work should include and what to ask before you sign anything. In some situations you'd be better off keeping your money, and we'll come to those as well.

What an AI consultant actually does

Good consulting starts with how the business runs today, well before anyone mentions a model or a tool. In a firm of accountants, that might mean following a client's documents from the moment they arrive by email to the moment they reach the bookkeeping software. In a lettings agency, it might mean listing every step between a viewing request and a signed tenancy. The consultant is looking for work that repeats, follows a recognisable pattern and eats into someone's day.

The useful output from that is short:

• a list of tasks, ranked by how much time they take and how often they come up
• a plain statement of what AI can and can't do on each one
• which applications would need connecting, and whether they allow it
• what data the AI would see, and what stays where it is
• how you'll know it worked, agreed before anything gets built

People skip that last point all the time, and it's the one that protects your budget. Hours saved, faster replies and enquiries processed can all be measured. "Becoming AI-ready" can't.

An AI implementation consultant goes a step further and stays through delivery: building the agent, connecting it to your CRM or calendar, testing it on real cases and showing your staff how to correct it when it slips up.

The questions to ask before you hire one

A handful of questions will separate the people who've done this from the people who've read about it.

What would you advise us not to do? If the answer is vague, or the plan touches every department, you're being sold scope. Anyone with experience will have a list of tasks they'd leave alone.

Can I see something you've built, running? Ask to watch a working system, and ask what went wrong with it along the way. Someone who really built it will have a story ready about what broke and how they fixed it.

What happens when the AI gets something wrong? It will. You want to hear about who reviews the output, what happens to the cases the system can't handle and how mistakes get fed back into it.

Who owns it when you've gone? Accounts, code, instructions and data should sit with you or be handed over. If only the consultant can change anything, you've bought a dependency.

Where does our data go? Anyone working as an AI consultant in the UK, or advising a UK organisation from elsewhere, should raise UK GDPR without being prompted: which suppliers handle personal data, under what agreements, and whether your customers' information can be used to train a model. That's a general pointer and not legal advice, so have your own adviser check anything specific to your organisation.

Do you earn anything from the tools you recommend? Referral fees aren't automatically a problem, as long as they're declared.

Advice only, or someone who also builds

There are two models, and each has a weakness.

A consultant who only advises hands you a report and you find someone else to carry it out. That can work well if you have developers in-house, or if a larger organisation wants an opinion from someone with nothing to sell afterwards. For a small business, the risk is paying twice, and then finding out during the build that the plan doesn't survive contact with your software. Plenty of recommendations fall apart the moment someone discovers the booking system has no way of letting anything else in.

A consultant who also builds will have tested their advice against what can really be made. The weakness is the obvious conflict of interest: someone who builds has every reason to recommend building. That's why the question about what not to do matters so much with this type.

We sit in the second camp, so read this section with that in mind. Our AI implementation service starts with how you work today, then moves on to building agents, connecting them to your software and teaching your team to use them. If you have a capable developer on staff and just want a second opinion, a pure adviser may suit you better.

When you don't need an AI consultant

Some of the most useful advice in this field costs nothing. You can probably skip the consultant if:

• You have one clear task and a product already does it. A salon whose booking software already sends reminders doesn't need anyone to analyse that.
• The task is rare. If it happens twice a month, automating it will cost more than the time it saves.
• There's no process yet. If every job is handled differently depending on who picks it up, write down how it should work first. Automation only makes an unclear process fail faster.
• You want to learn this yourself. A subscription to an AI assistant and a real task from your own working week will teach you a great deal. If you do come back to a consultant later, you'll be a much sharper buyer.
• You're hoping someone will promise that AI can run the place unattended. It can't, and anyone who says otherwise is worth avoiding.

What an AI consultant for small business costs, and why

We don't put prices in articles, because the honest answer depends on what's being asked for. What we can do is explain what the cost is made of:

• the time it takes to understand how you work
• how many applications have to be connected, and how awkward they are
• how much reading and interpreting the task requires, compared with simply moving data around
• the volume of work, since AI models and messaging or phone providers charge for use, and those charges carry on after launch
• training your team
• looking after the system once it's live

Treat a firm quote given before anyone has looked at your process as a warning sign. A better arrangement is to start with a single task at a fixed price, see the result, and only then decide whether to go further. That's how we work: you receive a fixed written price after the first conversation, before any work begins.

A sensible way to start

Pick the task that annoys you most. Write down how it gets done now, step by step, including the bits people do from memory. Then read about choosing the first process to automate, and if phone calls are the sore point, about what an AI receptionist can and can't do. Our AI implementation guide covers the stages in more detail.

If you'd then like a second pair of eyes, our team is based in Romania and works with companies abroad in English and German, so the first conversation happens on Zoom or by phone. We also use a chat agent and a phone agent in our own business, which means a fair amount of what we'd tell you comes from living with these systems. Book a conversation and bring your list. If the answer turns out to be that you don't need us, we'll say so.

Frequently asked questions