An AI consultant works out where AI and automation would help a business, then designs, builds or oversees the systems that do the work, with controls around them. A good one also tells you where AI is the wrong answer, and helps you pick a first project small enough to measure.
In short: an AI consultant maps how your business runs, picks the right tool for each problem (often plain automation, not AI), builds or oversees the system, and puts controls around it. Choose between an adviser, a builder, an agency and a platform vendor by whether you need direction, a working system, a larger programme or a packaged product.
The UK's Office for National Statistics reports that "In September 2026, 3 in 10 (30%) businesses reported use of at least one type of AI technology" (ONS, Business insights and impact on the UK economy, 8 October 2026). The consultant label now covers everyone from strategy firms that may not build anything to software vendors whose answer is often their own product.
What does an AI consultant actually do?
An AI consultant maps how your business runs, decides whether each problem needs AI, plain automation or a simpler process, then builds or oversees the system and puts controls around it. Those four parts usually run in that order, and you can check each with a supplier before you sign.
Understands how the business runs. Before any talk of tools, they map the processes that matter: how an enquiry becomes a customer, how an invoice becomes cash, how a management report reaches the people who decide. The aim is to find where time is lost, where errors creep in and where information is re-keyed by hand.
Picks the right tool for each problem. Many problems that arrive labelled "AI" are better solved with plain workflow automation, a tidier spreadsheet or a change to the process itself. AI is useful where the work involves reading unstructured information, such as emails, documents or phone calls, and making a first pass at it. A good consultant says which is which.
Builds, or oversees the build. Someone has to own the result: that it connects to your systems, copes with awkward cases and has a person checking the outputs that matter.
Puts controls around it. Any system that touches customer data, money or decisions needs to show what it did and why. That means logs, review steps, access limits and a named owner. Australia's privacy regulator, the OAIC, says that when a business adopts a commercial AI product, due diligence "should not amount to a 'set and forget' approach" (OAIC, Guidance on privacy and the use of commercially available AI products). The same thinking applies to anything a consultant builds for you.
What should an AI consultant not do?
- Start with the tool. If the first conversation is a demo of a platform before anyone has asked how your business works, the answer was chosen before the question.
- Promise outcomes nobody can measure yet. No one can honestly say how many hours a system saves before seeing the process. They can agree how it is measured, before and after.
- Treat data protection as your problem alone. The OAIC states that "the Privacy Act applies to all uses of AI involving personal information", and recommends that organisations do not enter personal information, and particularly sensitive information, into publicly available generative AI tools (OAIC). In the UK, the ICO publishes guidance on AI and data protection, which it notes is under review following the Data (Use and Access) Act. A consultant should design around the guidance in your country, while your legal adviser confirms what applies to you.
- Leave you with something nobody else can run. If only the consultant understands how it works, you have swapped one dependency for another.
Adviser, builder, agency or platform vendor: which kind of supplier do you need?
Each of these four is the right choice for some businesses.
| Supplier | What they do | Best when | Watch for |
|---|---|---|---|
| Adviser | Strategy, roadmaps, policies and training | You need direction and have a technical team to deliver it | Nobody inside the business can build what the roadmap describes |
| Builder | A consultant or small team that designs and builds the system | You know roughly what needs to change and want it working | Whether they also plan the controls and think about the people who use it |
| Agency | A larger team with designers, developers and project managers | You are running a bigger programme with several strands | Who actually does the work, and who looks after it once it is live |
| Platform vendor | A software company selling its own product | A packaged product fits your problem and you want to start quickly | Its advice points towards its own product |
The common mistake is hiring one kind when you need another, such as paying for a roadmap when you needed one process automated properly.
Where should a growing business start with AI?
Start with a process, not a technology. The best first candidates are frequent, rule-shaped and irritating: the same steps every time, information re-keyed between systems, a queue that builds up whenever someone is away.
Three questions narrow the list:
- Where do the hours go each week, and whose hours are they?
- Where do errors cost money, or embarrass you in front of customers?
- Which of those processes has a clear owner who would use a better system?
A process that scores well on all three, using data you may use for the purpose, beats the most impressive idea on the list.
Illustrative example, not a client. A wholesale business shortlists three ideas and scores each against the three questions.
| Candidate | Hours lost | Cost of errors | Clear owner | Verdict |
|---|---|---|---|---|
| Keying supplier invoices into the accounts | High, every day | Medium, wrong codes and duplicate payments | Yes, the finance lead | Start here |
| Replying to "where is my order?" emails | Medium, spread across the team | Medium, slow replies annoy customers | Partly, no single owner | Next, once it has an owner |
| A chatbot on the website | Low, few enquiries arrive there | Low | No | Leave for now |
The invoice process wins because it scores on all three and its data already sits in the business's systems. For a fuller method, see which business process to automate first.
In the UK, the government consulted on an AI governance self-assessment tool called AI Management Essentials. Its response says DSIT is not publishing the tool, and that future guidance is set to focus on foundational governance measures and specifically target SMEs (GOV.UK, AI Management Essentials tool: government response). Meanwhile, know which AI tools are in use, who owns each and what data goes into them.
What should you ask before you hire an AI consultant?
Ten questions to ask:
- What would you look at before recommending anything?
- Can you show us work you have built, and is it labelled as client work, an in-house tool or a demo?
- Who does the build, and who supports it afterwards?
- When would you tell us not to use AI?
- How do you decide between plain automation and AI for a given step?
- What controls do you put around a system that handles our data or money?
- Where does our data go, and which outside tools see it?
- How do we measure whether it worked, and what do we record before we start?
- What happens if the key person on your side is unavailable?
- Could our own team run and change this later?
Listen for specifics. Vague answers to questions 6 and 7 are the clearest warning sign.
What does a sensible first engagement look like?
A first project usually runs in four stages, whatever the technology.
Assess. A short look at your processes, systems and data, ending in a ranked list of opportunities and a list of things to leave alone.
Pick one process. Choose the one with the clearest owner and a starting point you can measure. Record how it works today: the time it takes, how often it goes wrong and where it gets stuck.
Build small. Build the smallest version that does real work, with a person reviewing its outputs at first. Connect it to the systems your team already uses.
Measure and decide. Compare the new process with the starting point. Then decide whether to extend it, change it or stop. Stopping is a valid result.
What drives the cost of an AI project?
No honest consultant can price a project unseen, but the drivers are predictable.
- Systems involved. Each connection to accounting software, a CRM or a phone system adds design and testing.
- The state of the data. Clean records are quicker to work with than years of free text and duplicates.
- Judgement. Rules-based steps are simpler to build and check than steps where the system interprets language.
- Control. Anything touching money, personal data or regulated activity needs more review, logging and testing.
- Hand-over. Passing it to your team needs documentation and training.
Ask any supplier to explain their estimate in these terms.
How do we work at GME Solutions?
We map how a process runs today, step by step, decide whether it needs AI, plain automation or a simpler process, and build around your own systems with the controls designed in from the start. We work with your data ourselves, or alongside your own data or tech team, and record the starting point so the result can be measured.
GME Solutions designs and builds custom AI solutions and workflow automation for growing businesses in Australia and the UK. The firm was founded by chartered accountant Greg East. His audit training taught him how businesses run, how to improve processes and how controls work, and every number we report traces back to its source. So we record who touches each step and which system holds each piece of information, and agree who reviews what and what gets logged.
If you want the wider plan first, see our guide to an AI roadmap for a small business. For examples, see our client builds: practice management for a UK accounting firm and live financial operations for an international events company. We are also building a custom voice agent for an insurance company, shaped by analysis of their historical calls.
If you have a process in mind, get in touch.
Frequently asked questions
What exactly does an AI consultant do? An AI consultant looks at how a business runs, finds the processes where AI or automation would help, and then designs, builds or oversees the systems that do the work. A good one also puts controls around those systems and tells you plainly when a simpler fix, or no AI at all, is the better answer.
How do you know if a business is ready for AI? You are ready to start when you can name one process that is frequent, follows clear steps and has an owner who wants it improved, and when you know what data it uses and whether you may use that data for the purpose.
How can AI automate business processes? Most business automation is rules-based: a trigger fires and a fixed sequence runs across your systems. AI adds value where a step needs reading or interpreting, such as sorting emails, pulling data from documents or summarising calls. Many builds combine both, with a person checking the outputs that matter.
Do AI consultants need to understand data protection? Yes. In Australia the OAIC states that the Privacy Act applies to all uses of AI involving personal information, and in the UK the ICO publishes guidance on AI and data protection. A consultant should design within that guidance from the start, while your own legal adviser confirms what applies to your business.
General information, written to be useful. It is not financial, tax, investment or legal advice. For decisions specific to your business, take advice from a suitably qualified professional.