The release of consumer health tools by OpenAI and Anthropic in recent weeks marks a significant shift in the global healthcare landscape. It signals that we are moving from the era of the portal to the era of the AI agent.
For the last decade, Australia’s digital health strategy has been sound. We focused on digitization and centralization. We built repositories like My Health Record and designed portals to display them. This was the correct approach for a document-centric world where the primary goal was ensuring a discharge summary could move from a hospital to a GP.
However, the technology landscape has evolved. The assumption that patients want to manage their own records by logging into a portal is being challenged. Recent developments suggest that patients do not necessarily want to manage records. They want answers to their health questions.
They are less interested in downloading a PDF of a lipid panel and more interested in knowing if their medication is working.
As 40 million people globally gain the ability to connect their health records directly to AI agents, we need to assess our national trajectory. We are currently optimizing for human readability in a world that is rapidly optimizing for machine readability.
To maintain our momentum, our strategy needs to expand.
The New Business Case for FHIR
The business case for FHIR has traditionally rested on interoperability. It was a long-term investment in moving data between organizations. This often made it difficult to prioritize against immediate operational needs.
The rise of AI agents changes this calculus.
When major AI companies launched their healthcare offerings, they built them on FHIR. This provides immediate validation for the investments Australian healthcare organizations are already making. The work being done on AU Core and through the Sparked accelerator is now doubly valuable. We are not just building compliance for national standards. We are building the necessary foundation to leverage the next generation of clinical AI tools.
The Importance of “Australia-Ready” Infrastructure
As we pivot to this new reality, we must remain grounded in technical detail. There is a significant difference between a platform that is “FHIR-compatible” and one that is “Australia-Ready.”
I often see organizations deploy platforms that claim global compatibility but struggle with local context. Australian healthcare has specific requirements. We have unique implementation guides, identifiers like Medicare and DVA numbers, and local clinical terminologies.
A generic FHIR server cannot always handle the nuance of an Indigenous status indicator or the specific workflow of an Australian eRequest. For AI agents to function safely in our system, the data must strictly conform to AU Core standards. Without this localization, the data lacks the semantic context required for safe analysis.
Evolving Our Strategic Direction
The rise of “Shadow Health IT” is a growing risk. If government portals remain static document viewers, citizens may start exporting their data to unsecured platforms to get the insights they want.
We do not need to discard our current roadmap, but we should look at where we can bridge the gap between our document-based present and the agent-based future.
Here are three practical adjustments we can consider:
1. Shift focus from UI to Infrastructure Our current strategy places a heavy emphasis on the front-end user experience, such as the 1800MEDICARE app. While important, the greater value for the government may lie in owning the trust layer. We should focus on the “plumbing” that allows a patient to securely connect the app of their choice to their data.
2. Formalize the Right to API Access Current regulations protect a patient’s right to access their data, which usually results in a document download. We should explore codifying a “Right to API Access.” This would mandate that healthcare providers offer a consumer-facing FHIR API, ensuring that when a patient connects to an AI agent, they do so through a regulated, standard channel.
3. expand Sparked to Consumer Scenarios The Sparked accelerator has been highly effective for provider-to-provider standards. We should consider expanding this to include consumer-to-agent workflows. We need to define the standards for how an AI agent authenticates and retrieves data on behalf of a patient within the Australian context.
Conclusion
The standalone Personal Health Record is not disappearing, but it is no longer the sole endpoint of digital health. We are moving toward Embedded Health, where data flows dynamically to where the patient needs it.
Australia has the right vision with the Health Connect Australia program. By making these strategic adjustments, we can ensure that the infrastructure we build today is ready for the reality of tomorrow.
Leave a Reply