Every healthcare AI discussion eventually turns to primary care. The potential seems obvious: GPs are under immense time pressure, managing complex patients, drowning in administrative burden. AI should help.

Yet AI adoption in Australian primary care remains limited, especially compared to hospital settings. Why?

I’ve spent time talking with GPs, practice managers, software vendors, and primary health networks about this. The barriers are structural, not just technological. Understanding them matters for anyone working in this space.

The Fragmentation Problem

Australian hospital care is delivered by relatively large organisations. A major metropolitan health service might have 10,000+ staff, substantial IT departments, and capacity to evaluate and implement new technologies.

Primary care is different. There are approximately 8,000 general practices in Australia. Most are small businesses with fewer than 10 GPs. Many are single-GP practices.

This fragmentation creates several problems for AI adoption:

No scale for implementation. Each practice would need to separately evaluate, procure, integrate, and manage AI systems. The overhead per practice is prohibitive for small operators.

Limited IT capability. Most practices don’t have dedicated IT staff. Technology decisions often fall to practice managers who are already stretched, or to GPs who’d rather be doing clinical work.

Varied technical infrastructure. Practices use different practice management systems, have different network setups, and have different levels of technical sophistication. AI products need to work across this heterogeneity.

No centralised coordination. Unlike hospital networks where a health service can mandate technology choices, no one can make practices adopt AI. Each practice is an independent decision-maker.

The Economics Don’t Work (Yet)

Hospital AI often has clear efficiency or revenue justifications. Faster radiology reporting means more throughput. Better coding means more accurate reimbursement. Reduced adverse events means lower liability costs.

Primary care economics are different:

Time-based billing constraints. Under Medicare, GPs bill for consultations, largely time-based. AI that makes a consultation faster doesn’t necessarily translate to more revenue—it might just mean shorter consultations for the same billing item.

No AI-specific MBS items. There are no Medicare rebates specifically for AI-assisted care. Until AI use creates billable events, the financial case is harder.

Upfront costs, uncertain returns. AI products have licensing costs. Integration and training cost time. The financial return is uncertain and often distant. For practices operating on thin margins, this is a difficult investment.

Productivity vs profitability gap. Even if AI makes GPs more efficient, converting that efficiency to revenue isn’t straightforward. You can’t see 40 patients an hour even if AI theoretically enables it.

The Workflow Reality

Primary care workflows are remarkably compressed. A typical GP consultation is 12-15 minutes. In that time, the GP needs to:

• Understand the presenting problem
• Review relevant history
• Examine as needed
• Formulate and communicate a plan
• Document the encounter
• Manage prescriptions, referrals, follow-up

Any AI intervention needs to fit into this compressed timeframe without adding friction. AI that requires additional clicks, screens, or attention competes with patient care time.

The most promising AI applications for primary care are those that:

• Work in the background without requiring attention
• Integrate into existing workflows rather than creating new ones
• Reduce documentation burden rather than adding to it
• Provide value within the consultation timeframe

Ambient documentation—AI that listens to consultations and drafts notes—fits these criteria better than many other AI applications. That’s why it’s getting more traction than diagnostic AI in primary care.

The Vendor Challenge

The primary care software market in Australia is dominated by a handful of practice management systems: Best Practice, Medical Director, and a few others. These systems handle patient records, billing, prescribing, and most clinical workflows.

For AI to work in primary care, it generally needs to integrate with these systems. But:

Integration is technically complex. Each practice management system has different architecture, different data models, different integration capabilities.

Vendor incentives are mixed. Practice management vendors may see AI integration as a feature they should provide themselves, not something they should enable competitors to provide.

Data access is complicated. AI often needs access to clinical data, raising privacy and data governance questions that practice management vendors may be cautious about.

Market size limits investment. The Australian primary care market, while significant, isn’t large enough to attract the same vendor investment as larger international markets.

What’s Actually Working

Despite barriers, some AI applications are gaining traction:

Ambient documentation tools. Products that record and transcribe consultations are appearing in more practices. The value proposition—reduced documentation time—is clear and immediate.

Practice management AI features. Some practice management systems are building AI features into their products: automated appointment reminders, intelligent scheduling, clinical coding suggestions. This approach avoids integration complexity.

Pathology and radiology AI (indirect). GPs benefit from AI used in pathology and radiology even if they don’t use AI directly. Faster, more accurate results improve GP decision-making.

After-hours triage. AI-supported triage for after-hours services and nurse-led care lines. This is adjacent to general practice but affects primary care patients.

What Could Accelerate Adoption

Several changes could accelerate AI adoption in primary care:

MBS reform. If Medicare recognises AI-assisted care in billing structures, the economics change. This doesn’t mean extra payments for using AI—it might mean recognising AI as a component of care that enables different service models.

Primary health network coordination. PHNs could play a greater role in evaluating and coordinating AI adoption across practices. Shared services, shared procurement, shared support could address scale barriers.

Practice management system integration. If major practice management systems built AI capabilities natively or enabled easy third-party integration, the technical barriers would reduce.

GP-led innovation. Some of the most successful primary care innovation comes from GPs themselves identifying problems and working with developers to solve them. Supporting GP-led innovation matters.

For practices that are ready to explore AI, working with knowledgeable partners helps navigate the options. AI consultants Sydney and similar organisations increasingly work with primary care settings, though the implementation model differs from hospital deployments.

My Assessment

I’m genuinely uncertain about the timeline for significant AI adoption in Australian primary care. The barriers are real and structural.

Conversations with AI consultants Brisbane who work with both hospital and primary care clients confirm this picture—the implementation model that works in hospitals simply doesn’t translate to fragmented primary care.

My best guess:

Short term (1-2 years). Ambient documentation gains traction in practices willing to invest. Practice management systems add incremental AI features. Most practices continue without significant AI.

Medium term (3-5 years). Integration options improve. Economics become clearer as evidence accumulates. Early majority practices begin adopting. PHN coordination increases.

Longer term (5+ years). AI becomes expected infrastructure for primary care, like electronic prescribing or My Health Record. Practices without AI are at competitive disadvantage.

But this timeline assumes steady progress without major setbacks. A significant AI-related adverse event in primary care could slow adoption. Regulatory changes could accelerate or decelerate. Economic pressures on general practice could crowd out innovation investment.

What GPs Should Do Now

For GPs and practice managers thinking about AI:

Focus on documentation first. Ambient documentation is the most mature primary care AI application with the clearest value proposition. If you’re going to try AI, start here.

Wait for your practice management system. If your practice management vendor is developing AI features, using those is likely easier than integrating third-party products.

Talk to your PHN. Some PHNs are exploring coordinated AI initiatives. Find out what’s available in your region.

Don’t feel behind. Most practices aren’t using AI yet. You’re not missing a revolution. When AI becomes clearly valuable and practical for primary care, adoption will be more obvious.

The potential for AI to help Australian general practice is real. The timeline for realising that potential is longer than enthusiasts suggest. That’s not defeatism—it’s realism about the structural changes required.

---

Dr. Rebecca Liu is a health informatics specialist and former Chief Clinical Information Officer. She advises healthcare organisations on clinical AI strategy and implementation.