AI Consultant for Small Business: What One Actually Does, What It Costs, and How to Choose

A small business does not need an AI strategy. It needs the three or four jobs that eat the owner's week taken off them, reliably, without a new hire. An AI consultant worth paying is the person who works out which jobs those are, specifies what should do them, and either builds it or stands over the build until it works. Everything else in the title is marketing.

What an AI consultant does, step by step

1. Maps where the hours go. Not a survey. A walk through the actual week: the inbox, the chasing, the reporting, the onboarding, the meetings written up at nine at night. Each recurring task gets a number: minutes each, times a month, who does it.
2. Scores what to automate. Time saved against risk of getting it wrong. Reporting scores high (lots of hours, low risk). Anything touching money or clients scores lower on risk and needs a person in the loop. The order comes out of the scores, not out of what is fashionable.
3. Specifies the build. Which systems it reads from, what it writes, where a human approves, what happens when it fails. A consultant who cannot write this is going to hand you a deck.
4. Builds it, or stands over the build. The good ones build. The honest ones who do not build bring their own developer and stay accountable for the result.
5. Proves the hours back. Before and after, in numbers. If the consultant cannot show you the hours reclaimed, you cannot tell whether it worked.

The four kinds of AI consultant you will meet

• The strategist. Workshops, roadmaps, a 40-page document. Useful for a 200-person company deciding policy. For a 12-person business, the roadmap is the easy part and nobody builds it.
• The tool reseller. Certified in one platform, recommends that platform. Fine if the platform happens to fit; expensive if it does not, because you will be running your old process inside a new subscription.
• The automation freelancer. Builds one workflow in Make or Zapier, invoices, disappears. The workflow breaks in month three and nobody knows how it worked.
• The operator-builder. Has run a business, can do the operations thinking and the technical build, and stays to run what they built. Rarest and the only kind that gives a small business its hours back for good.

That last description is what Anco AI is built to be. Dave Crawford ran agencies before this one and runs Anco AI on the same AI Operating System he installs for clients: the invoices, the bookkeeping, the proposals, the daily brief are all automated in his own business first. The AI consultancy page sets out what that looks like.

What it costs

Published prices are rare in this market, which is itself a warning sign. Ours, for comparison:

• Discovery call: free, 30 minutes. You leave with a map of where the hours go whether or not we work together.
• Scaleability Audit: £997. A 48-hour deep dive and a prioritised action plan you own.
• A single build (an AI email responder, automated onboarding, a reporting system): typically £2,500 to £15,000 depending on what it touches. An AI enquiry responder for a premium service business was £3,000; meeting-intelligence software for an accountancy firm was £7,500.
• Ops and tech consulting retainer: from £3,500 a month, three-month minimum, for a business that wants the thinking and the building on tap.
• AIOS Foundation: £4,995, then £995 a month to host, monitor and add an automation every month.

The AI automation cost guide goes into what moves a quote up or down.

How to choose one in a single call

Five questions. The answers tell you which of the four kinds you are talking to.

1. "What did you build last month, and what does it do now?" A builder answers with specifics. A strategist answers with a framework.
2. "Show me it running." Anyone who has built real systems can share a screen. Ours is the business you are talking to.
3. "Where does a person approve things in what you'd build for us?" If the answer is "nowhere, it's fully automated", walk away. Money and client communications need a human gate.
4. "What happens when it breaks at 7am on a Monday?" You want a monitoring answer and a named person, not "it won't".
5. "What will you tell me not to automate?" The good ones have a list. Automating a broken process gives you a faster broken process.

What a first engagement looks like, week by week

• Week 1, the audit. Two sessions with the owner and whoever runs operations. Every recurring task listed with minutes, frequency and owner; every system named; every place the owner is the bottleneck marked. Output: a scored list and a recommendation for the first build.
• Week 2, the specification. One page per build: what it reads, what it writes, where a person approves, what happens when it fails, what "done" means in hours saved. Priced. Nothing is built until this is agreed.
• Weeks 3 and 4, the build. On the real systems with real data, with the owner seeing a working slice by the end of week 3. The human approval step is built first, not bolted on.
• Week 5, the measurement. Hours before against hours after, in the owner's own numbers. If the number is not there, the build is not finished.

That is the shape of every engagement we run, from a £3,000 email responder to a full operating system. The timetable changes; the sequence does not.

Three mistakes small businesses make when hiring one

1. Buying the roadmap. A strategy document for a 12-person business is a list of things nobody has time to build. Pay for the first build, not for the plan to build.
2. Letting the consultant choose the tool first. If the recommendation arrives before the audit, it is the tool the consultant knows, not the tool you need.
3. Automating the founder's favourite task. The task that annoys the owner most is rarely the one that costs the most hours. The audit exists to stop that argument with numbers.

If you want those five answered about your own business, that is what the discovery call is for.

Dave Crawford

Founder, Anco AI. Builds the systems this site talks about, and runs his own business on them.