What AI consulting actually costs a small business

In short: Almost every firm in this category will not put a number on it until the third meeting. We put one on it in the first thirty minutes. Here is what you are actually paying for, what moves the number up and down, the second bill that arrives after ours, and what a suspiciously cheap quote is really buying you.

Key takeaways

• Four things get sold under one name, and they do not cost the same: an audit, a build, tool selection, and operating what got built. Which one you need is the first thing worth settling, and it is free to settle.
• A build is typically 10 to 30 hours of senior engineering time. A quote that implies far less than that is selling you a demo, not a deployment.
• There is a second bill. The software your system runs on is not part of the consulting fee, it renews every month, and it is the number most proposals leave out.
• The biggest driver of a project going over is not technical difficulty. It is being scoped without full visibility into how the operation actually runs.

What you are actually paying for

Four different things get sold under one name. They are priced differently because they are different work, and knowing which one you need is most of the way to knowing what it should cost.

Discovery call. Free, thirty minutes, always. You get a number on this call, not in a proposal three days later. If we get off it and the honest answer is that you should not buy anything from us yet, that is the answer you get.

AI readiness audit. A fixed fee, agreed before it starts. We map how your operation actually runs, identify five to ten places where AI would take real work off your team, and hand back a prioritized plan with what each one would cost to build. You own the document. You are free to take it to another firm or build it in house. There is a longer piece on what an audit actually answers.

Custom build. A fixed fee, quoted against the audit, once the work is actually known rather than guessed at. A single workflow end to end is the cheap end. Work that spans several systems costs more, because the thing has to talk to your CRM, your accounting software, and your phone line, and each of those connections is its own small project.

Tool selection and implementation. Less than a build, because nothing gets written from scratch. You do not need something custom. You need someone to pick the right off the shelf product, configure it properly, connect it to what you already run, and train your team on it.

Voice agents. A setup fee, then a flat monthly fee. This is our most requested build and the one with the clearest before and after. Flat is the part worth noticing: the human answering services meter by the minute or by the call, so their bill goes up exactly when your business gets busy.

Operating what got built. A monthly fee, cancel any month. Not maintenance in the sense of waiting for something to break. Someone continuing to build, monitoring what is running, and fixing the things you would not have caught.

Every custom build has both a setup fee and a monthly fee. If a proposal has a setup fee and no monthly fee, ask what happens when the API on the other end changes. Something on the other end always changes.

What moves the number

Three things, in the order they matter.

How many systems the work has to cross. One workflow inside one piece of software is the cheap end. The phone line into the platform, the platform into the CRM, the CRM into the reporting, that is three projects wearing one name.

How much of your data has to be untangled first. Nobody quotes for this and it is the most common reason a build goes long. If the addresses are in four formats and half the contacts are duplicates, that gets fixed before anything intelligent runs on top of it.

Whether anyone can tell you what actually happens. The scoping cost is lower in an operation where somebody can walk you through a Tuesday. It is higher where the answer is "it depends who is on."

Why an audit costs what it does

Because the work underneath it is ten to thirty hours of a senior engineer's time, and that engineer is the same person who has to build the thing afterward.

Do the division yourself, on any quote you get, ours included. Take the number, divide it by a plausible hourly rate for a senior engineer, and see how many hours it implies. When someone quotes an audit at a price that works out to two or three hours, they are not more efficient than us. They are running your business description through a language model and sending you the output. You can do that yourself in an afternoon and keep the money.

The reason the hours are what they are is that most of an audit is not technical. It is sitting with the person who actually does the work and watching what they do, because what people describe when you ask them how a process works and what actually happens on a Tuesday are two different processes. The gap between those two is where every real automation opportunity lives, and you cannot find it from a questionnaire.

The second bill

This is the part that gets left out of proposals, including, historically, some of ours.

Our fee is for the work. The system we build for you runs on software that bills you separately, every month, forever. Depending on what we build, that can include an automation platform, a model API that charges per use, a phone carrier, a database, and sometimes a seat on a product you did not previously pay for.

None of that is enormous. All of it is real, and it does not stop. Before you sign anything with anyone, ask for the monthly running cost of the system separately from the monthly fee for the humans, and ask what happens to that number if your volume doubles. Usage based pricing is fine. Usage based pricing you find out about in month four is not.

The honest framing is that the monthly fee on a proposal is never the monthly cost. It is that fee plus the running costs underneath it, and the person selling it to you should be able to tell you both numbers before you sign.

What a cheap quote is buying you

We lose deals on price. Sometimes that is the right outcome, because there are businesses that should be spending a couple hundred dollars a month on a good off the shelf product instead of many thousands with us, and when that is true I will say it on the discovery call.

But the cheap quotes that go wrong go wrong in a predictable way. Someone builds the happy path. It works in the demo. Then it meets your actual inbox, where a large share of the messages are formatted differently than the sample they tested on, and the system starts producing confident wrong answers that nobody catches for weeks. We have written about why most AI pilots die after the demo, and the short version is that the cases the system gets wrong are the actual job. The cases it gets right are the demo. Pricing that only covers the demo is not cheaper. It is a deposit on doing the whole thing again.

The specific questions to ask any quote, ours included:

1. How many hours of engineering time is this, and who is doing them?
2. What happens to my system when the software on the other end changes?
3. What is my monthly running cost, separate from your fee?
4. Can my team turn it off without calling you?
5. What is the plan for the cases it gets wrong?

A firm that cannot answer question five has not thought about your operation. They have thought about the model.

What actually moves the price

Not complexity. Visibility.

The engagements of ours that have gone over on time or scope share one thing. We priced them without full sight of how the operation really ran. Someone described a process, we scoped against the description, and then during the build we found the four exceptions that everyone in the business handles by habit and nobody thinks to mention. Those exceptions are not edge cases to the business. They are Tuesday. And they are half the build.

That is the argument for the audit as a separate paid engagement rather than free pre sales work. Not because we want another line item, but because a build priced on top of a real map of your operation is priced correctly, and a build priced on top of a conversation is a guess. If you take the audit and go elsewhere with it, you will still get a better quote from whoever you hire, because they will be quoting against something real. That sequencing is the whole reason we work in audit, build, operate rather than quoting a build off a first call.

What to do with this

1. Get the number on the first call. If a firm will not give you a band before a proposal, that is information about how they work.
2. Ask for the running cost separately from the fee. Both numbers, in writing, before you sign.
3. Divide the quote by a plausible number of engineering hours. If the result is implausible, you are buying a demo.
4. Pay for the map before you pay for the build. A scoped build is priced. An unscoped build is a guess, and guesses go over.
5. If your honest read is that you are too small for any of this right now, that is a real answer and the right firm will tell you so.