AI for small business in Australia: where to start

Most advice about AI for small business is written by someone selling a tool. This guide is built from what Australian government sources actually measured and recommend, because those are the only numbers here that nobody is paid to inflate. The short version: AI pays in a small business when it is aimed at one measured task, gated by a person, and kept away from customer data until the settings are checked.

How many small businesses actually use AI

You will see two very different numbers quoted, and both are correct.

The Australian Bureau of Statistics reported in June 2026 that around 12 percent of Australian businesses used AI in 2024-25, up from 1 percent in 2021-22. The gap by size is wide. Around 35 percent of large businesses and 22 percent of medium businesses used AI, while small and micro businesses sat at around 11 percent. Adoption also tracked innovation closely: small businesses that were actively innovating used AI at 19 percent, against 4 percent for those that were not.

The National AI Centre runs a separate monthly survey of small and medium enterprises. Across December 2025 to February 2026 it found 43 percent reported some level of AI adoption, rebounding to 44 percent in February. That survey counts any level of use, including limited one-off experiments, so it sits far above the ABS figure.

Read together, the picture is simple. A lot of small businesses have tried AI. Far fewer have built it into how the business runs. That second group is the one seeing a return, and the National AI Centre notes that businesses already using AI are shifting from one-off use toward broader use across the business. Once a use case proves itself, owners tend to expand it rather than retreat.

Why most small businesses have not started

The same National AI Centre survey asked non-adopters why they did not intend to implement AI in the next 12 months. Three reasons dominated.

Trust. Around 65 percent of non-adopting businesses cited distrust of AI decision-making or a strong preference to keep human control of their processes. That is a reasonable position, and it is also the easiest one to design around. Nothing in a sensible first project requires handing a decision to a machine.

Relevance. 54 percent said AI is not relevant to their business. Adoption in construction and agriculture sat below 30 percent, while health, education and services were above half. The survey’s own reading is that this reflects a lack of relatable examples rather than a rejection of the technology.

Not knowing where to start. 19 percent said they simply do not know how to use AI in their business, up two points on the previous quarter.

If you recognise your business in any of those three, the rest of this guide is aimed squarely at you. We cover the trust objection in more depth in is AI worth it for a small business.

Where AI pays in a small business

Among Australian SMEs using or planning to use AI, the National AI Centre found content generation and data analytics were the most common applications, each at 54 percent, followed by cybersecurity and threat detection at 48 percent. More complex uses, such as agentic AI, supply chain optimisation and AI-assisted hiring, remain largely untapped.

That ordering is sensible and worth copying. The uses that pay first share three traits: the work is repetitive, it is mostly text or numbers, and a person already reviews the result. In practice that looks like:

• Drafting. Customer email replies, quotes, product descriptions, job ads and social posts drafted by AI and edited by a person before they go out.
• Summarising. Long supplier contracts, meeting notes, call recordings and customer reviews condensed into the three points that matter.
• Reporting. Sales, stock and marketing numbers pulled together into a weekly summary instead of a Friday afternoon spent in spreadsheets.
• Answering routine questions. Opening hours, delivery times and returns handled from your own documented policies, with anything unusual passed to a person.
• Finding the leaks. Abandoned checkouts, unanswered enquiries and overdue invoices surfaced automatically so someone can act on them.

Notice what is not on the list: anything where the AI acts alone on money, customers or legal commitments. That boundary is not caution for its own sake. It is what lets a first project succeed while the trust question is still open.

The safety rules worth following first

Three free Australian government resources cover most of what a small business needs, and they agree with each other.

The privacy regulator. The Office of the Australian Information Commissioner recommends, as best practice, that organisations do not enter personal information, and particularly sensitive information, into publicly available generative AI tools. The Privacy Act applies to any use of AI involving personal information, whatever the tool.

The Australian Cyber Security Centre. Its guidance for small business, written with New Zealand’s National Cyber Security Centre and the Council of Small Business Organisations Australia, names three risks: data leaks, unreliable or manipulated outputs, and supply chain weaknesses. It warns that some AI providers can use submitted data to train their models depending on settings and subscription type, and it recommends an internal AI use policy that clearly defines what data cannot be uploaded, staff training on checking outputs, and removing or anonymising personal details before using an AI tool.

Business.gov.au. The government’s business portal recommends identifying the problem before choosing a tool, involving staff early, and piloting AI in one or two areas before a broader rollout. It points to two National AI Centre templates that are worth downloading on day one: an AI policy template and an AI systems register for keeping track of which tools the business uses and for what.

None of this requires a lawyer or a security team. An afternoon with those two templates puts a small business ahead of most of its competitors on governance.

A first project you can actually measure

The most common reason AI spending disappoints is that nobody wrote down what it was supposed to change. Here is a sequence that avoids that.

1. Pick one task. Choose something that costs real hours every week and already gets checked by a person. Quoting, customer email and weekly reporting are typical candidates.
2. Time it now. Record how long the task takes over two normal weeks. Without this baseline you will never know whether the tool paid for itself.
3. Set the rules. Fill in the AI policy template, decide what data is off limits, and turn off model training in the tool’s settings where the option exists.
4. Run it for a month. Use the tool on that one task only. A person approves every output before it reaches a customer.
5. Measure and decide. Compare the hours against your baseline and count the corrections you had to make. Keep it, change it, or drop it. A tool that saves no time is a subscription, not a system.

Only once one task is paying back is it worth adding a second, or connecting the tool to your other systems. That connection work is where projects get harder, and where AI integration services start to matter.

When outside help is worth it

A single tool on a single task does not need a consultant. Most owners can run the five steps above with free government guidance and a month of attention.

Outside help earns its place in three situations. The first is when the work crosses several systems, such as a store, a CRM and an accounting package, and the data has to move reliably between them. The second is when customer data is involved and you want the setup documented rather than improvised. The third is when you have already tried a handful of tools and genuinely cannot tell whether any of them is working, which is exactly the question an AI maturity assessment is designed to answer.

If that describes your business, our AI foundations work covers training, use case ranking and tool selection before anything is built, and our AI automation agency service builds and measures the automations that come out of it, in accounts you own. The AI agency page shows how the two fit together. For a view of where AI would pay in your business specifically, the free AI use-case audit ranks the candidates by hours saved and revenue impact, in plain English, with no obligation attached. Victorian businesses can see how the same work runs locally on the AI consultant Melbourne page.