What does an AI consultant do, and when you do not need one

An AI consultant looks at how your company works day to day, finds where AI or automation could save time or reduce mistakes, and helps your team put it into use with agreed data rules, human checks and tests. For a business of 10 to 100 people that usually means a written plan first, then one or two working setups, and rules for what data may go into AI tools.

That is the useful version. There is also a version where you pay for a slide deck full of trends, so it is worth knowing what to expect and when to say no.

What does an AI consultant do in practice?

In my work the job has four parts, and a good consultant will tell you which of them you need.

1. Take a picture of how things work now, from facts rather than impressions. Most owners know their company by its big results, and the details where the time goes are often invisible from the owner's desk: who retypes the same order into two systems, how long an enquiry sits before anyone answers, which reports are put together by hand every Friday. Before anyone talks about AI, someone has to write this down.

2. Choose what to do first, and what not to do. A list of fifty AI ideas is worthless. What you need is two or three tasks ranked by effect and effort, with the risks of each, and a short list of things that look tempting but should wait. Sometimes the best first step has no AI in it at all, only a form that sends enquiries straight into the customer database.

3. Set it up so it is harder for people, or the AI, to get it wrong. In practice it means agreeing what company data may go into which AI tool, who checks AI output before it reaches a customer, and how the setup behaves when something fails. It also means accepting that not everyone in the team is comfortable with computers, and designing for that instead of hoping for it.

4. Hand it over. Training, a short written description of how it works, and someone in your team who owns it. Ideally your employees should feel the change as less manual work inside the tools they already use, not as a new program they have to learn.

Some consultants stop after step 2 and hand you a plan. Others, like me, also do steps 3 and 4. Both are fine, as long as you know which one you are buying.

What should you have at the end?

Something you can check. After a first consultation that means a written plan for one area: the tasks worth starting with, the effort and risk of each, what not to do yet, and what the first step would cost. After a setup it means a working flow you have tested yourself with a test enquiry or a test invoice, plus the list of subscriptions it runs on and who pays for them.

If you cannot tell after the work what has changed in your company, the work was advice, and advice should be priced as advice. More on that in how much an AI consultant costs.

When you do not need an AI consultant

Here it is plainly. You probably do not need one if:

• Your question is about one tool you already pay for. If the question is "how do I make our email tool sort messages by topic", the vendor's help pages or support team will answer it faster and cheaper than I will.
• The problem is a process nobody owns. If enquiries get lost because no one is responsible for answering them, AI will lose them faster. Decide who owns the process first; the tools come second.
• Nobody will have time to use the result. A setup without an owner in your team turns into another unused subscription within a few months. If the next quarter is already full of other work, wait.
• Your volume is small. If you receive only a few enquiries a week, compare the time spent handling them with the setup and monthly costs. A simple manual process may be enough.
• You need a software product built. A custom app for your customers is a job for a development team. A consultant can help you write down what you need, but should not pretend to be the team.

None of these is a failure. They are good reasons to keep your money.

Why so many businesses are unsure whether they need AI at all

You are not alone in not knowing where AI fits. In a UK government survey of around 3,500 private businesses with at least five employees, the most frequently cited reasons for not adopting AI were "a lack of identified need and limited AI skills and expertise", and around one in six businesses (16%) were using at least one AI technology (DSIT, AI adoption research, page updated 13 February 2026). "We do not see where we would use it" is a normal starting point, and finding the answer to it is most of what a first consultation is for.

From our own system: what a plan looks like, and the rule I work by

I have published a sample AI roadmap for a fictional cleaning company of 32 people in the UK, marked as a sample, so you can see the format before you pay for anything: the tasks, the effort and risk of each, and what to leave for later.

The rule I work by on every free call: afterwards you get either a short written outline with up to three tasks worth starting with, or "you do not need AI for this". I say the second one when it is true.

I also run my own work on a system of AI agents (research, documents, report drafts) and a person approves the templates, rules and company knowledge it works from and checks samples of what goes out, which is the same rule I set up for clients.

How to choose an AI consultant

Ask three questions on the first call. What will I have at the end, in writing? What is not included? What would make you tell me not to do this? A consultant with clear answers to all three is worth a second conversation. If you want the next step after that, there is a practical outline in how to implement AI in a small business: a 30-day plan, and for the rules side, an AI acceptable use policy your team will read. Other first-step guides are in getting started with AI.

Not sure you need one?

That is exactly what the free call is for. In 30 minutes we look at your situation, and afterwards you get a short written outline with up to three tasks, or an honest "not now". If you want a full written plan for one area, that is the paid consultation, with the plan delivered within 24 hours.

Sources

• UK Department for Science, Innovation and Technology, AI adoption research, gov.uk, page updated 13 February 2026, accessed 18 October 2026.
• Elivance consultation, elivance.eu/services/consultation (our own page). Prices from the Elivance service specification dated 17 October 2026; excl. VAT.