What an AI consultant for a small business actually does, and how to pick one

Why the title tells you nothing

Anyone can call themselves an AI consultant, and in 2026 most of them sell the easy part. The Federal Reserve's 2026 Report on Employer Firms, from a survey of 6,525 small employer firms, found 46 percent of small businesses using AI. Among those users, about half were still experimenting, 44 percent had partially integrated AI into a process, and 7 percent had fully integrated it. The most common uses were writing or marketing at 83 percent and individual productivity at 61 percent.

That is the experiment stage, and it does not need a consultant. A team can open a chatbot and draft an email without help. The paid work starts at the integration step, where the tool has to read from one system, write to another, and stop when the input is wrong. That step is also where the failures live. RAND interviewed 65 data scientists and engineers and found that more than 80 percent of AI projects fail, twice the rate of IT projects without AI. The causes it names are not technical: the business and the builders never agree on the problem, the data the project needs does not exist, the team chases the technology instead of the problem, and the infrastructure to run the result is missing.

MIT's NANDA initiative reached the same place from the other direction. Its 2025 report on generative AI in business, from 150 interviews, a survey of 350 employees, and 300 public deployments, found that about 5 percent of AI pilots produced rapid revenue gains and the rest stalled. The reason was a learning gap, not model quality: generic tools do not learn a workflow. One more finding matters for anyone hiring: buying from a specialized vendor or partner succeeded about 67 percent of the time, and internal builds succeeded a third as often.

So the job of an AI consultant is to get a small business through the integration step without becoming one of those numbers. Everything below is that job broken into its parts.

The four jobs, and what each costs

Every real engagement we have seen or run is one or more of these four. A consultant who cannot tell you which one they are selling is selling hours.

1. Find the process

The first job is choosing where AI pays. The consultant looks at how the business runs, finds the work that repeats every week and loses time, fees, or accuracy between systems, and ranks it. The output is a plan with a first move, not a slide deck about the future of AI. Published discovery engagements run $2,000 to $5,000 and end in a written plan (Layer3 Labs). Our AI Audit is this job over two weeks with a fixed scope, ending in a ranked plan: what to start, what to defer, what to leave alone.

2. Choose the tools

The second job is finding out whether an existing product already covers the process, and getting the team to use it. This is where most first projects should end, because a tool that fits costs less than anything built and can be running this month. Our AI Finder does the shortlist for free, and Buy Your Time Back does the whole job for $995: a 45-minute look at a normal week, a path built from existing tools, and a check that the team got five hours back.

3. Build the integration

The third job exists when no product covers the work, which usually means the work is hard in the gap between two systems that no vendor owns. The consultant builds the piece that moves data across that gap, with rules for what leaves the building and what the system refuses to decide. A defined implementation for a small business runs $5,000 to $25,000, with $10,000 to $15,000 typical (AIessentials), and reaching into a legacy system pushes it toward the top of that band (Layer3 Labs). Our dental payment integration is this job: recurring payments moved to ACH without replacing Dentrix, with patient matching and ledger posting kept inside the clinic, and every uncertain match stopped for a person.

4. Run it afterward

The fourth job starts the day the system goes live. Exceptions surface, the software it touches gets updated, and the next process is waiting. Some businesses take this in-house with a named owner. Others keep the consultant on a retainer, which published tiers put at $2,000 to $8,000 a month (AIessentials). Either way, the question to settle before the build is who does this job and at what cost. Our AI Partnership is the retainer version, and we hand the controls to the team in every case.

The full price bands for each job, with sources, are in their own article.

Which job you need

Three questions sort it, and you can answer them before the first call.

• Can you name the process? If you can describe one task that repeats every week, arrives in the same shape, and has one owner, skip the finding job and go to tools. If you cannot, the finding job comes first, and adding a tool before it makes the mess faster.
• Can you price it? Staff hours, fees, errors, or delay. A process with a number attached is a project. One without is a conversation.
• Does a product cover it? Check before anyone builds. If one tool handles the work without rearranging your workflow, buy it. The build job is for the gap no product owns.

Most small businesses land on the first two jobs. The build job is real, and it is rarer than the people selling it suggest.

Seven questions for the first call

Each of these has a short right answer. A long answer to any of them is the answer.

1. Show me one system you built that is running in production today. Not a demo. A record: what it does, what it touches, and what it refuses to do.
2. What data would leave my building, and where would it go? The answer should be one sentence you could repeat to a customer.
3. What is the price tied to, and what does it exclude? A scope, or a process, or a result. Data cleanup, licenses, and maintenance named as numbers or as zeros.
4. Who owns the system after go-live? A name on your side, a path for exceptions, and a monthly cost for the consultant's part if there is one.
5. What happens when the input is ambiguous? The right answer is that the system stops and hands the case to a person. A system that guesses is unsupervised, not confident.
6. Which of my existing software do you keep? A consultant who needs you to replace your practice management system, your accounting package, and your CRM before the first process is selling a migration.
7. What would make you tell me not to do this? Every honest consultant has turned work away. Ask for the last time.

Five red flags

• The platform comes before the process. If the recommendation arrives before anyone has looked at a normal week, it was decided before you called.
• Every case study has a multiplier and no detail. Ten times faster, ninety percent cheaper, no client, no system, no date. A real record names the constraint and says what the evidence does not support.
• They want all your data first. A first project needs the data for one process. A request for everything is a request to figure out the business on your time.
• Hourly with no scope. The hour is the wrong unit for a first project. A consultant who cannot bound the work cannot price it, and the number will move.
• The demo is the product. If the answer to question one is a demo, there is no production system. Demos are cheap. Tuesdays are expensive.

Does the consultant need to be near you

"AI consultant near me" is one of the most common ways people search for this, and location is the wrong filter. The finding job needs to see how the work moves, which a screen share and an hour with the person who does it cover. The build job needs access to your systems, which is a credential, not a drive. What matters is that the consultant can observe the process repeatedly and reach your team during your working hours.

Where being nearby helps is the discovery day for a physical operation: a clinic, a warehouse, a shop floor. If your work happens on a floor rather than in software, ask whether the consultant will stand on it once.

What we do

We do the first three jobs and offer the fourth as a partnership. The AI Audit finds the process. Buy Your Time Back chooses the tools and is refunded if the team does not get five hours back in a normal week. A custom AI system builds the integration when no tool covers the gap. Our case studies state only what the available record supports, and where it does not support a number, the page says so instead of claiming one.

The question we ask on every first call is the one we would want asked of us: what does this process cost you today? If nobody knows, that is the first job, and it does not require hiring anyone yet.

Frequently asked questions

Questions owners ask us

What does an AI consultant do for a small business?

Four things: finds the process where AI pays, picks existing tools that fit it, builds a custom integration when no tool covers the work, and keeps the system running afterward. Most small businesses need the first two jobs first. A consultant should be able to say which of the four they are selling and what it costs.

How do I find an AI consultant near me?

Search by the process rather than the postcode. The work happens over screen shares and system access, so a consultant two time zones away who can observe your process repeatedly beats a local one who cannot. Being nearby matters for a discovery day at a physical operation such as a clinic or warehouse. Ask for a production system, a data boundary, and a scoped price before asking where they are.

What is the difference between an AI consultant and an AI agency?

A consultant advises and often builds one process at a time, usually with the person you spoke to doing the work. An agency sells a packaged service, often on a platform it resells, delivered by a team you meet after signing. Neither label guarantees anything. Ask both the same seven questions and hire whoever answers them in one sentence each.

How much does an AI consultant cost for a small business?

Published 2026 rates put independent consultants at about $150 to $350 an hour, discovery engagements at $2,000 to $5,000, a defined first project at $5,000 to $25,000, and retainers at $2,000 to $8,000 a month (AIessentials, Layer3 Labs). Our entry engagement is $995, payable only after we confirm fit. The full cost breakdown has every band with its source.

What should I ask an AI consultant before hiring?

Ask to see one system running in production, what data would leave your building, what the price is tied to and what it excludes, who owns the system after go-live, what happens when an input is ambiguous, which of your existing software they keep, and when they last told a client not to proceed. Each has a one-sentence right answer.

Do I need an AI consultant, or can I do it myself?

You can do the finding job yourself if you can name one weekly process, price what it costs today, and check whether an existing tool covers it. Our ten-question diagnostic and AI Finder are free for exactly that. Hire when the work crosses systems no product connects, or when nobody on the team has time to own the change.

What this is based on

Sources

Every figure above is quoted from a published source or from work we have shipped. Checked September 5, 2026.

1. Federal Reserve Banks, "2026 Report on Employer Firms". Findings from the 2025 Small Business Credit Survey, fielded September 3 to November 14, 2025, 6,525 responses. AI use, integration depth, and most common uses.
2. RAND Corporation, "The Root Causes of Failure for Artificial Intelligence Projects and How They Can Succeed". Interviews with 65 data scientists and engineers; the 80 percent failure rate and its causes.
3. Fortune, "MIT report: 95% of generative AI pilots at companies are failing". August 18, 2025. Coverage of MIT NANDA's "The GenAI Divide: State of AI in Business 2025", including the vendor-versus-internal success rates.
4. Layer3 Labs, "AI Consulting Rates and Pricing in 2026". Updated June 17, 2026. Discovery and project bands.
5. AIessentials, "AI Consultant Cost (2026)". Updated August 21, 2026. Hourly, project, and retainer bands.
6. Dental payment integration and the evidence standard our case studies follow.