Plenty of firms now call themselves AI consultants. Some are strategy advisers, some build software, some resell a platform, and some do a little of each. For an owner trying to put AI to work in an established business, the label tells you very little. What matters is what they do, who they answer to, and what you hold when they leave.
This guide sets out what a good AI consultant should do, the warning signs to watch for, and the questions to ask before you sign anything.
What an AI consultant should actually do
A good AI consultant starts with the business problem, not the technology. Before anyone talks about tools, they should be asking where your team's time goes, which problems cost the business the most, and whether AI is the right answer at all. Sometimes a process change, or better use of the software you already own, does the job.
From there, the work usually covers:
- Opportunity assessment. Where time is going, which problems matter and whether AI is the appropriate answer.
- Vendor and proposal review. Questioning the scope, dependencies, data access and ongoing costs of the options in front of you.
- Priorities and a roadmap. What should happen first, what can wait and what needs more investigation before anyone spends on it.
- Implementation planning. The requirements and decisions to settle before committing to a build, so any provider can quote against the same brief.
If the consultant also builds, the build should follow the same order: map the work, agree the design, build and test it with the people who use it, then hand it over with documentation.
Warning signs to watch for
- The answer arrives before the questions. If the recommendation is a particular platform before anyone has looked at how your business runs, the advice is probably a sales pitch.
- Promises about outcomes. Promised savings, fixed percentages or claims that a system is always accurate. AI drafts, sorts and checks well, but a person should still approve anything that commits the business.
- A product to sell. Advice tied to seats, licences or a platform's sales targets is not independent, even when it is good.
- Vague ownership. If it is unclear who owns what gets built, the accounts it runs on and the documentation, you may be locked in.
- Nothing about your data. A proposal that doesn't say what data the system needs, where it is processed and who can access it is incomplete.
- Fees with no scope. You should know the deliverables, responsibilities, timing and fees before work begins.
Questions to ask before you sign
- What business problem does this solve, and how will we know it has worked?
- Have you looked at whether a process change or our existing software would do the job?
- Which of our current tools will this connect to, and what, if anything, do we have to replace?
- Which steps run automatically, and where does a person review or approve the output?
- What data does it need, where is that data processed and who can access it?
- Who owns what gets built, the accounts it runs on and the documentation?
- What happens if the vendor changes its pricing, or we want to leave?
- What does it need from our team to keep running, and what support is included?
- Do you sell or resell any of the software you are recommending?
- What exactly is in the scope, and what are the fees?
A good consultant will be comfortable answering all of these in writing.
Consultant, software vendor or in-house hire?
A software vendor makes sense when you already know the problem and an off-the-shelf product fits it well. The trade-off is that the vendor's advice will lead to its own product.
An in-house hire makes sense when AI work will be continuous and you have enough of it to keep a person busy. The trade-off is finding, managing and keeping that person, and giving them the authority to change how the business runs.
A consultant makes sense when you need an independent view, or a working system built into your operation, without building a team first. The trade-off is that you depend on how clearly the scope, ownership and handover are agreed.
These are not exclusive. Many businesses get independent advice first, then build with whoever fits the job.
How we work
Turner & Barnett is a consulting firm, not a software vendor. We don't sell seats or licences. Our independent AI advisory stands on its own; you do not need to commission a build. When you are ready to build, our AI implementation work connects the tools worth keeping and builds the workflow around how your team operates. There is no rate card; read what an engagement costs for how fees are set, or meet the founders on our about page.
If you run a building or trades business, see AI for construction businesses in Australia.
Want a second opinion on an AI quote?
If you have a proposal on the table and want an independent view before you commit, book an introductory call or start a discovery brief.