What AI automation actually costs a small business

The most common question we get is some version of “what would that cost?”, and the most common experience people have had before asking us is that nobody will tell them.

That is not usually evasiveness. Scopes genuinely vary. But it leaves business owners unable to answer the only question that matters, which is whether the thing is worth doing at all. So here are actual numbers.

Two caveats before the figures. These are guides and starting points, not a rate card, and a real quote follows a look at your actual setup. And they are what we charge in Melbourne in 2026, which is a specific market at a specific time.

There are two costs, and people only think about one

The mistake almost everyone makes is treating automation as a single purchase, like buying a laptop.

It is two things. There is the build, which is one-off, and there is the running cost, which is forever. Wildly different in size, and the second one is the one that surprises people, because software has trained us all to think in flat monthly subscriptions.

AI does not bill like that. It bills per use. We wrote about this when Canva slowed its own AI rollout over per-task costs, and the same mechanic applies at your scale. Every draft, every summary, every document read has a cost, and the bill grows with use.

The good news for a small business is that at your volumes, the meter is usually small. It is the build that dominates.

What a build costs

Roughly, in bands:

Ongoing care, where someone keeps the thing running, monitors it and fixes what breaks, starts around $450 a month for the basics and runs to roughly $850 a month when it includes reporting and continuous improvement.

If you only take one thing from that table, make it this: the useful stuff starts in the low thousands, not the tens of thousands. The pitch that automation is an enterprise-scale investment is mostly people describing enterprise-scale projects.

What it costs to run

The meter is the part nobody quotes, so here is how to think about it.

Cost per run, multiplied by runs per month. That is it.

A short task, drafting a follow-up email or summarising an enquiry, costs cents. A business sending 200 of those a month is spending single-digit to low-double-digit dollars. A task that reads long documents, like drafting from a 40-page tender, costs meaningfully more per run, but you also do far fewer of them.

Add whatever subscriptions the automation sits on, a CRM seat or similar, and for most small businesses the whole running cost lands somewhere between a phone bill and a modest software subscription. It is real, it should be in your numbers, and it is rarely the thing that kills the business case.

Three worked examples

A trades business chasing quotes. Build a quote and booking flow that sends the follow-ups automatically. Call it $2,400 to build, and $20 to $50 a month to run once you count the tooling.

The way to test it: if your average job is $1,200, the build pays for itself the moment it recovers two jobs you would otherwise have lost to silence. Most businesses lose more than two a year to nobody following up. That is a straightforward yes.

A clinic or services firm losing enquiries. Enquiries arrive by form, email, phone and Instagram, and land in four places. A growth systems setup puts them in one list with automatic follow-up and actual reporting, at $3,500 to $6,500.

Harder to put a single number on, because the gain is enquiries that stop falling through gaps. Which is why we would want to know your enquiry volume and your conversion rate before quoting, and why you should be suspicious of anyone who does not ask.

A professional services firm writing proposals. An AI assistant drafting from your own past work and knowledge base, from $4,500.

The honest measure here is hours, not dollars. If a proposal currently takes six hours and drops to two, and you write four a month, that is sixteen hours back every month. Multiply by what an hour of your senior people is worth and you will know quickly whether it stacks up. We are not going to do that multiplication for you, because the number is different in every business and inflated ROI claims are how this industry lost its credibility.

Doing several at once. An established business tackling three of these together typically lands around $8,800 one-off, plus care from roughly $850 a month. That is the shape of a genuine first phase rather than a dabble.

What makes a quote go up

Not the AI. Almost never the AI. What drives cost is the state of what it has to connect to:

• Your data is in five places. Most of the work is reconciling it, not automating it.
• There is no single source of truth. If two systems disagree about a customer, someone has to decide which wins, and that is a business decision that takes meetings.
• Something has no API. Legacy or niche software that will not talk to anything turns a simple job into a custom one.
• You want it unsupervised from day one. Building something that runs with a human checking it is far cheaper than building something trusted to run alone, because the second one needs to handle every edge case properly.

This is the same point we made about why the barrier to AI adoption is the plumbing, not the tools. The technology is cheap. Connecting it to a real business is the work.

How to spend less

• Start with one task. The most annoying repetitive job in your week. One, done properly, beats five half-built.
• Tidy your data first, even manually. An afternoon deduplicating a customer list can take real money off a quote.
• Accept a human in the loop at first. Let it draft, you approve. Cheaper to build, and you find out whether you trust it before it matters.
• Do not automate a broken process. Fix the process, then automate it, or you will pay to make a bad thing happen faster. This is the whole argument in fixing the flow before you buy the tool.

The func.digital take

The reason we will put numbers in public is that the alternative wastes everyone’s time. Owners who assume this costs $50,000 never ask. Owners who assume it costs $500 are disappointed. Both would have been better served by a table.

The honest summary: for most Melbourne small businesses, the first genuinely useful automation costs somewhere between two and five thousand dollars to build and less than a phone bill to run, and it either pays for itself in recovered work or it does not. That is a question you can answer on paper before you spend anything, and if the answer is no, we would rather tell you that than sell you something.

If you want that answered for your business specifically, the digital systems audit is free for businesses we can genuinely help. We will look at what you are running, tell you which one thing is worth automating first, and put a number on it. Get in touch, and let’s find out whether the maths works.