THE EIGHTEEN-MONTH GAP
Walk into ten allied health clinics in any Australian city today and ask what AI they’re using in their operations. You’ll get the same answer in eight or nine of them: nothing. Not “we’re piloting something,” not “we tried it, and it didn’t work.” Nothing.
That same week, you can sit down with operators in Texas, in London, in Toronto, and watch them route inbound calls through voice agents, run patient reactivation cycles off automated sequences, and pull operational summaries out of a dashboard that’s been running for a year. The technology isn’t different across borders. The vendor lists are different. The compliance environments are different. But what’s actually different, when you look honestly, is time. Australian businesses are roughly eighteen months behind their international peers on AI adoption. Not behind in capability. Behind in deployment.
I want to write about why, and what I think the path through it looks like, because I’m spending all of my days inside that gap.
Here’s what the gap is made of.
The first thing is that most of the AI tools Australian operators see pitched to them aren’t built for Australia. They’re US tools that assume US infrastructure, US compliance, and US customer expectations. When an operator runs a procurement check, one of two things happens. The tool fails the Privacy Act 1988 review, or the vendor can’t explain how it doesn’t. Either way, the operator walks. Quietly. Without telling anyone, they walked, because nobody’s keeping score. So the operator is left thinking AI doesn’t work in their industry, when in fact the imported version doesn’t work in their jurisdiction.
The second thing is that the people best positioned to build the local versions are usually too busy. A clinic owner running four practitioners and a reception team isn’t going to spend their Tuesday night learning prompt engineering. A medium-sized accounting firm in regional Victoria isn’t going to fund an internal AI engineering function. So the work that needs doing, the translation work, sits unowned. Generic horizontal tools fill the vacuum (the chatbots, the meeting transcribers, the generic AI writers), and operators try them, get nothing meaningful, and conclude AI is overhyped. Which is it, horizontally? But vertically, industry by industry, with someone willing to sit inside a single sector long enough to build what actually fits, it isn’t overhyped at all. It’s just unbuilt.
The third thing is conviction. The operators I’ve spoken to don’t lack interest. They lack a credible deployment partner. Most have already been pitched by someone overseas, or by an Australian agency reselling a US stack, and the experience left them wary. The next time someone shows up with “AI for your business,” they default to no. Which is fair. The way the category has been sold has earned that no.
So the question I keep coming back to is what closes the gap.
The honest answer, the one I’d write down if I were starting from scratch, is that the gap doesn’t close horizontally. There isn’t a single Australian AI platform that’s going to ship and lift all industries at once. Generic AI doesn’t translate. It has to be translated. The translation has to be done by someone willing to sit inside one industry, long enough to understand what an operator actually does at 7:42 on a Tuesday morning when the receptionist hasn’t shown up and three patients are waiting on hold. Then build the thing that fits that specific moment. Then deploy it. Then watch it. Then improve it. Then move to the next industry and do it again.
That’s what Clearing Bars is doing. Currently inside allied health clinics. Chiropractic, Chinese medicine, osteopathy, physiotherapy. The first deployment of Pulse, the system I built for clinic operators, has been live at Auburn Wellness Centre since the start of the year. Fifteen missed calls were recovered in week two. Around ten hours a week of admin is handed back to the team. Privacy Act and AHPRA compliant by design. No imported tooling that doesn’t fit.
What I want to do with Field notes is write down what’s getting learned inside that work, in long-form, so other operators (and any other founder who decides to take a single Australian industry through the same translation) can see what holds up and what doesn’t.
A few of the pieces I expect to publish over the coming months. How to actually scope an AI deployment that survives contact with reality. What the Privacy Act and AHPRA shortlists really look like once you’ve done the work. Why most clinic AI tools you see advertised are solving the wrong problem. What the second industry will be and why. What an Australian operator should ask any vendor before they sign anything. Where AI is genuinely returning hours, and where it’s still a slide deck dressed up as a product.
I want to keep these pieces long-form for a specific reason. The category has already been hollowed out by short-form. Threads, hot takes, eight-bullet listicles. Most of what’s been written about AI for Australian businesses is reactive, optimised for clicks, and forgettable. The operators who are actually weighing a deployment decision don’t need another thread. They need the thinking behind the thinking. They need to see how someone closer to the work would reason through their problem. That’s what I’m trying to write.
If that’s useful to you, subscribe. The cadence will be roughly fortnightly. Considered, not constant. I’d rather skip a week than ship a piece that wasn’t ready, and I’d rather write three good essays a quarter than thirty disposable ones.
If you run an Australian business and you’d like the translation done inside your industry next, write to me. Some of the most interesting essays here will come out of conversations like that.
The eighteen-month gap closes for whoever does the work to close it. I’m starting with one industry. Field notes are where I’ll write down what I find. See you in the next one.
Paulo