Good AI consultant services end with a working system your team runs and owns, the logins to every piece of it, and a document someone else could maintain it from. If the engagement ends with a strategy deck and a list of software to subscribe to, you bought advice. You still have to build the thing, and you're paying for the tools while you figure out how.
TL;DR
Ask for this list in writing before any work starts. If a consultant won't commit to it, you find out before you've paid for anything.
Turn that list into the scope of work, with a date on each line. Then ask to see a maintenance document from another project. A consultant who has handed off real systems should have one. The work itself should move through four stages from first call to handoff, and each line of the scope should map to one of them.
It's how AI Marketing Labs scopes Custom AI Automations: we build on your accounts and tools, test against the actual work, train your team, and hand it over with no ongoing fee to us.
Because the plan is the easy part, and a tool that sits beside the work is easy to ignore. MIT's Project NANDA reviewed more than 300 publicly disclosed AI initiatives from the first half of 2025 and found that only 5% of integrated AI pilots were extracting millions in value, while the vast majority showed no measurable impact on the P&L. The report names a core reason: most of these systems "do not retain feedback, adapt to context, or improve over time."
That research looked at enterprise AI. A ten-person shop has less room for a pilot that doesn't ship, which is the reason to put the handoff in the contract instead of hoping for it at the end.
Price follows how much actually gets built. A May 2026 career guide on the University of Miami's career site lists typical ranges: strategy assessments at $5,000 to $25,000, workflow implementation projects at $5,000 to $75,000, team training at $2,000 to $15,000, and fractional AI officer retainers at $3,000 to $15,000 a month (How to Become an AI Consultant). Treat those as one guide's read on the market, not quotes.
Inside those ranges, the price moves with how many of your systems the workflow has to touch, how messy the data going into it is, and how many people need training to run it. A build that reads one inbox and writes to one spreadsheet costs less than one that pulls from your CRM, your accounting software, and a shared drive.
The number to push on is what you pay after the engagement. A system stitched together from subscriptions the consultant picked can cost more each year than the build did. Ask for the monthly cost of every tool in the proposal, and ask what happens to the system if you stop paying the consultant.
A proposal tells you what the consultant thinks they're selling. Read it for four things.
You can learn a lot in thirty minutes if you listen for what the consultant asks about. A good one spends the call on your process: who touches the work, where it waits, what it costs you when it's late. Be wary when the call goes the other way.
None of these prove bad intent. They do tell you the engagement is set up to keep you coming back rather than to leave you running on your own. Before you sign with anyone, vet the firm on what it has already built: call a reference who uses the system every day, and read the contract for who owns your data.
When you don't know which problem to solve yet. If your honest answer to "what should AI do for us" is "a bit of everything," a short assessment is worth more than a build, because a build would aim at the wrong target. Keep that phase small and fixed-price, and make its output one named bottleneck with a rough cost to fix it. A roadmap of twelve initiatives is a sign the assessment went too wide.
A quick way to tell whether you need an AI strategy consultant or someone who builds: if you can name the process, who does it, and how many hours it eats, skip the strategy. And if you're choosing between an AI specialist and a general consultant who has added AI, pick by the problem, not the title.
If you're still deciding whether you need outside help at all, start with when to hire an AI consultant for a small business. And if you already paid for a strategy that ended as a deck, hand it to whoever builds and ask which single item they'd ship first.
Pay a fixed price for a scoped build, because you know what you're getting and the consultant carries the risk of going over. Hourly makes sense for open-ended advice, like a few working sessions to decide what to build first.
What's the difference between an AI consultant and an AI agency?An AI consultant can be one person who advises and sometimes builds. An agency is a team that builds and may also maintain systems for you. Either can work. What matters is whether you own the result and can run it without them.
What happens if the consultant disappears after launch?If the accounts are in your name and the maintenance document is any good, another builder can pick it up. If the accounts are in the consultant's name, you may have to rebuild. That's why ownership belongs in the contract, not in a handshake.
Can an AI consultant work with the software we already use?A good one should build on it. Ask which of your current tools they plan to connect, which new subscriptions they'd add, and why each one is needed.