Artificial intelligence (AI) can now write large amounts of working software. Tools such as Claude Code, OpenAI Codex, Replit and Lovable allow developers to describe what they want in plain English and generate functioning applications in minutes. 

For the technology industry, this moment is often described with a blunt phrase: AI is eating software. But in healthcare, the implications may be even more interesting. 

It may change who builds healthcare technology in the first place. I spent more than fifteen years working as a physiotherapist before retraining as a software engineer. 

One of the most surprising discoveries during that transition was how healthcare software is actually built. Most systems are created by talented engineering teams who work hard to understand clinical environments from the outside. 

They interview clinicians, observe practice workflows and attempt to translate those insights into digital tools. But the reality of clinical practice is nuanced. 

A small workflow decision in software can add minutes of friction to every consultation. Documentation templates can subtly influence clinical thinking. 

Administrative processes can shape how a clinic operates day to day. In other words, the details matter enormously. 

And many of those details are only visible to the clinicians working inside the workflow. 

Over the past few years, I have had the opportunity to work on both sides of this intersection – practising clinically while also building healthcare software. 

What becomes clear very quickly is that many of the most valuable product insights do not emerge from technology teams alone. 

They emerge from clinicians describing the small frictions they experience every day: the referral letter that takes too long to write, the documentation template that interrupts clinical flow or the patient information that is difficult to summarise.

These problems may seem small individually but solving them well requires deep understanding of clinical practice. For decades, building healthcare software required significant technical resources. 

A typical development cycle looked something like this: an idea emerged; a software company hired engineers; months or years of development followed; clinics then adapted their workflows to fit the finished system. 

This model produced many of the digital systems used across healthcare today, including practice management platforms, electronic health records and scheduling tools. 

However, it also created a gap between the people designing the technology and the clinicians using it. AI is beginning to change this equation. 

Modern AI coding systems can generate application logic, write tests, suggest improvements to existing systems and help developers build software far more quickly than before. 

Instead of starting with thousands of lines of code, developers can now begin with a simple instruction: 

‘Build a physiotherapy documentation tool that converts consultations into structured clinical notes.’ 

Within minutes, a functional prototype can begin to take shape. 

Experienced engineers are still essential for building secure and scalable healthcare systems. 

Clinical software requires strong architecture, privacy protections and rigorous testing. But the distance between an idea and a working prototype has dramatically collapsed. 

What once required a full development team can now often begin with a small group of people – sometimes even one. 

This technological shift highlights an important reality. The scarce resource in healthcare innovation is not coding skill. It is clinical insight. 

Physiotherapists, occupational therapists and other allied health professionals spend thousands of hours inside consultations. 

They see where information gets lost, where administrative friction occurs and where systems fail to support clinical reasoning. 

These problems are often invisible to external developers. 

Historically, clinicians had to communicate these issues to software teams and hope they were implemented correctly.

 AI-assisted development changes that dynamic. 

Clinicians with ideas for improving healthcare workflows can now participate far more directly in building solutions. 

Early prototypes can be created quickly, tested in real environments and refined based on clinical feedback. 

The result is a new possibility: clinicians becoming builders of healthcare technology. 

For many years, clinicians have primarily been users of software. 

Systems were purchased, implemented and adapted to existing workflows. 

But as the tools required to build software become more accessible, clinicians may increasingly become designers of the systems they use. 

A physiotherapy clinic that identifies friction in documentation, patient education or reporting workflows may no longer need to wait years for a vendor to implement a feature request. 

Instead, those ideas can be explored, tested and refined much more rapidly. 

In some cases, this may lead to clinician-founded startups. In other cases, it may simply lead to better internal tools that improve patient care and reduce administrative burden. 

Either way, the balance of influence may begin to shift. 

Healthcare technology is not the same as consumer software. Any system used in clinical environments must operate within strict regulatory, legal and professional frameworks. 

In Australia, digital health tools that influence diagnosis, clinical decision-making or treatment may fall under the oversight of the Therapeutic Goods Administration as ‘software as a medical device’. 

Depending on how a system functions, this may require registration, risk classification and ongoing regulatory compliance. 

Even when a tool does not meet the threshold of a regulated medical device, healthcare technologies must still comply with Australian privacy and security requirements, including obligations under the Privacy Act 1988 and guidance from the Office of the Australian Information Commissioner. 

Patient data storage, encryption, access control and cloud infrastructure must all be carefully designed to protect sensitive health information. Professional regulation is also important. 

The Australian Health Practitioner Regulation Agency has issued guidance on the use of AI in healthcare, reminding practitioners that AI tools should support – not replace – clinical judgement. Clinicians remain responsible for decisions made in patient care, even when digital systems assist documentation or analysis. 

Transparency and consent also remain central to ethical practice. 

Patients should understand when AI tools are involved in documentation or workflow support and clinicians must ensure that these technologies align with professional and ethical standards. 

In other words, while AI may make software easier to build, it does not remove the responsibility to build and use it safely. Healthcare innovation must move forward but it must do so thoughtfully. 

AI will not replace physiotherapists. Clinical reasoning, therapeutic relationships and patient trust remain at the heart of healthcare. But AI may fundamentally change how the tools surrounding clinical care are created. 

For the first time, the ability to build healthcare technology is becoming accessible to the people who understand healthcare problems best. The next generation of healthcare software may not come exclusively from large technology companies. Some of it may come directly from clinicians themselves. 

For much of the past 30 years, clinicians have adapted to software. In the next decade, software may finally adapt to clinicians. For allied health professionals, this shift presents a meaningful opportunity. 

Clinicians who understand both the realities of practice and the possibilities of emerging technology are uniquely positioned to shape the future of healthcare systems. 

Not by replacing engineers, but by working alongside them. 

Because when clinical insight meets modern software tools, something powerful can happen. Physiotherapists may finally be in a position to build the systems they work with every day. 

Barry Nguyen APAM is a physiotherapist, a software engineer and the founder of CliniScribe AI. He is a member of the APA’s AI Advisory Group and a digital health adviser for the Australian Digital Health Agency.