Parliament wants to know what is stopping small businesses using AI. We can save them some time

The newly formed Joint Select Committee on Artificial Intelligence has opened an inquiry into why Australian businesses, and smaller ones in particular, are not adopting AI faster. It reports on 30 November, and its remit is broad: sovereign AI capability, worker displacement and job security, AI-powered scams, and how intellectual property law copes with any of it, reported by SmartCompany.

The adoption question is the one worth engaging with, because a lot of the submissions will get it wrong in the same predictable way.

The usual answers are cost, skills and awareness. Businesses do not know what AI can do, cannot afford it, and have nobody who understands it. Those sound right, and they are what a committee will hear over and over. But look at what has actually been built in Australia over the past year and the story stops holding together.

Access is not the problem anymore

There is a free government hotline you can ring for AI advice, which we wrote about in July. There is subsidised help through the Digital Solutions program. There is a National AI Plan. The models themselves cost less than a phone plan, and the free tiers are better than what enterprises were paying for two years ago.

Awareness is not the barrier either. Every business owner in the country has heard of ChatGPT. Most have used it. We covered the survey data showing most Australian SMEs have adopted AI while almost none have changed how they work, and that gap is the actual story.

So the raw materials are cheap, the advice is free, and the awareness is universal. Adoption still stalls. Something else is going on.

The research the committee should not skip

Here is the uncomfortable part. An OpenAI working paper cited in the inquiry coverage suggests early adoption of enterprise-grade tools “may exacerbate the performance gap between large and small businesses”, because larger firms adopt sooner and are better placed to integrate the technology properly.

Read that carefully, because it inverts the hopeful version of this story. The technology that was supposed to let a small firm punch above its weight may be doing the opposite. Not because big companies have better AI, they do not, everyone has the same models. Because big companies have someone whose actual job is wiring it into the systems they already run.

That is the barrier. Not the tool. The wiring.

What the barrier looks like on a Tuesday

In practice, in Melbourne, it looks like this.

Your customer data is spread across an inbox, a spreadsheet, a booking system and someone’s head. Enquiries arrive by email, by web form, by Instagram DM and by phone, and they live in four different places with no single list. You have no CRM, or you have one nobody has updated since 2024. Your quoting tool does not talk to your accounting package, so someone re-keys the numbers. Your job sheets are a shared document that half the team edits and half the team prints.

AI cannot fix any of that by itself, and pointing a chatbot at a mess produces a confident mess. The work that makes AI useful is unglamorous: getting your data into one place, making processes explicit enough that software can follow them, and connecting tools that currently only connect through a person copying and pasting.

A large firm has a team for that. A twelve-person clinic or building company has an owner who could maybe look at it on a Tuesday night, between quoting and payroll. That, not cost or awareness, is the gap.

Three things worth doing before November

None of these need a transformation program or a consultant with a deck.

1. Get every enquiry into one list. Web form, email, phone, socials, all landing in the same place with a name, a date and a status. This is the foundation for everything else, and on its own it usually finds work you were quietly losing.
2. Pick the one task your team complains about most and automate that. Quote follow-ups, appointment reminders, chasing missing information. One task, measured properly, beats a grand plan that never ships.
3. Connect two systems that currently talk through a human. Whatever your worst copy-and-paste job is, that is the one. Every hour of re-keying you remove is an hour back permanently.

Do those three and you have done more than most businesses will do this year, without buying anything exotic.

What we would tell the committee

If someone asked us to sum it up in a sentence: small businesses do not need more AI advice, cheaper models or another awareness campaign. They need help doing the plumbing, and there is currently nobody funded or incentivised to do it at their scale.

Grants aimed at buying technology will keep missing, because the technology is already cheap. Support aimed at integration work, the unglamorous business of getting one business’s data and processes into a shape software can use, would land. That is a less exciting thing to announce, which is probably why it does not get announced.

The func.digital take

There is a version of the next two years where the productivity gains from AI accrue almost entirely to businesses that already had good systems, and everyone else falls further behind while paying for subscriptions they barely use. The research suggests that is the default path, not the pessimistic one.

Avoiding it does not require getting ahead of the technology. It requires getting your own house in order first: one list of customers, processes clear enough to hand to software, and tools that talk to each other. Do that and today’s AI is genuinely useful to you. Skip it and next year’s will not help either.

If you want a straight assessment of which parts of your operation are ready for this and which are held together by someone’s memory, that is what a free digital systems audit is for. No transformation program, just a clear list of what to fix first. Get in touch, and let’s do the plumbing.