The appeal is obvious. You have a pathology report written for a doctor, not for you, full of abbreviations and reference ranges. You would like to ask a plain question - what does this actually mean, is it different from last year, should I be asking someone about it - and get a plain answer.
In 2026 that is finally a real product category rather than a demo. It is also moving fast enough that most of what you will read about it online is already out of date, and the biggest two options do not work in this country. This article was checked on 30 July 2026, and it says what is available in Australia and what is not.
Disclosure
I founded exora, which does this. So I am describing a category I compete in, and you should weigh what I say accordingly.
The approach I have taken: state what each product does and where it is available, source it, and be specific about what mine does not do. I have also included the option most people actually use - pasting documents into an ordinary chatbot - and treated it seriously rather than dismissing it, because it works and because dismissing it would be transparently self-serving.
The short answer
- ChatGPT Health is real, became broadly available in late July 2026, and is United States only. An Australian account cannot connect medical records to it.
- Perplexity Health is real, launched March 2026, and is also US only, on the paid Pro and Max tiers.
- Pasting your documents into ordinary ChatGPT, Claude or Gemini works from Australia right now, and is what most people do. It is genuinely useful for understanding jargon. Read the three caveats below before you do it with your whole history.
- Apps you can actually use here that have this built in include Tidy Health PHR on iPhone, and exora on iOS, Android and web.
- The one thing worth judging any of them on: can you check the answer against the document it came from, in one step? If not, you are trusting fluency.
ChatGPT Health
OpenAI announced Health in ChatGPT in January 2026, tested it with a limited group, and made it broadly available in late July 2026. It lets you connect Apple Health and supported medical records so that ChatGPT can draw on your medications, lab results, recent visits, sleep and activity data inside a normal conversation - explaining a result, summarising what has changed, or helping you prepare for an appointment.
On the record side it connects through the Epic and Oracle Health electronic record systems, and named participating providers including One Medical and Function Health. It is available to logged-in users aged 18 and over on the Free, Go, Plus and Pro plans, on web and iOS.
OpenAI’s stated privacy position is that connected health information and the conversations that use it are not used to train its models or to target advertising, and that connected health information receives additional encryption. OpenAI also says it works with a network of physician advisers across dozens of countries on evaluating model performance, and states plainly that the feature “is intended to support, not replace, professional medical care”.
Availability: the United States. As at 30 July 2026 there is no Australian rollout. (Source: Digital Health News, 30 July 2026; OpenAI’s own announcement is at openai.com.)
Perplexity Health
Perplexity launched Perplexity Health in March 2026. It pulls together medical records, lab results and connected device data into one place, answers health questions against your own history, and provides a dashboard that trends metrics over time. It connects Apple Health, electronic health records from a large US provider network, and wearables including Fitbit, Ultrahuman and Withings.
Perplexity’s stated position is that health data connected through the feature is encrypted in transit and at rest, is not used to train its models, and is not sold to third parties. Its answers carry citations to clinical guidelines and research, and include guidance on when to seek professional care - which, as an approach to a health question, is the right instinct.
Availability: the United States, on the paid Pro and Max tiers. (Source: TechRepublic, 23 March 2026.)
Why “US only” is a bigger deal than it sounds
It would be easy to read those two availability notes as a rollout queue - Americans first, Australians in a few months. It is worth understanding why that is not necessarily how this goes.
The hard part of these products is not the conversation. It is the connection to your records, and that connection is built on national healthcare plumbing. In the United States, consumers can authorise access to their hospital records through the patient-portal interfaces of the big electronic record vendors, and a whole industry of intermediaries exists to broker it. That is what ChatGPT Health and Perplexity Health are plugged into. Australia has no consumer-facing equivalent of that layer. What we have is My Health Record, and as at July 2026 no consumer AI assistant connects to it.
This is the same wall Apple hit. Apple’s Health Records feature - the one that downloads your clinical records from a provider into the Health app - is listed by Apple as available in Canada, the United Kingdom and the United States only, checked 30 July 2026. It launched in the US in 2018. Australians have been waiting eight years.
So the honest expectation to set is: these are good products, they may well come here, and nobody should hold their breath. In the meantime the gap is filled by whatever you can do with the documents you already have.
What Australians can actually do today
Option 1: paste documents into an ordinary AI assistant
ChatGPT, Claude and Gemini are all available in Australia, all accept an uploaded PDF or a photograph, and all will do a genuinely good job of explaining what a discharge summary means in plain English. This is the most common thing people do and it would be dishonest of me to pretend it has no value. If your goal is “translate this paragraph of clinical shorthand”, it is hard to beat and it is free.
Three caveats, in order of how much they matter.
It has no memory of your history. Each conversation starts from nothing except what you paste into it. That is fine for one document and useless for the questions that actually matter over time - is this trending up, when did this first appear, have I been on this medication before, does this result differ from the same test two years ago at a different lab. Answering those requires a record, not a chat.
You cannot easily check the answer. More on this below, because it is the central issue.
Your health information leaves the country, and may be used for training unless you say otherwise. These are US-based services, so uploading your records means processing them overseas under another country’s law. Separately, general-purpose assistants typically have a setting governing whether your conversations are used to improve the model - in ChatGPT it lives under Settings, then Data Controls. Find it and decide deliberately rather than by default. Two things worth knowing: turning it off is not retroactive, and a training opt-out is not deletion, since providers generally retain conversations for a period for safety and abuse monitoring regardless. Check the current terms of whichever service you use, because these settings change and any article’s snapshot goes stale.
Option 2: apps available here with it built in
Tidy Health PHR is on the Australian App Store, free with in-app purchases, iPhone and iPad only. It is a personal health record that keeps data on your own devices and markets an AI feature answering questions from your own record. The on-device approach is a real privacy advantage; the iPhone-only limitation is a real one if anyone in your household is on Android.
exora is mine, covered below.
I compared these and the rest of the field, including the free government options, in a wider comparison of health record apps in Australia.
The thing actually worth judging these on
Every product in this article will give you a confident, fluent, well-organised answer. That is what these models are good at. The problem is that a fluent wrong answer and a fluent right answer look exactly the same, and in a health context you are the least equipped person in the room to tell them apart.
There is a useful illustration of the shape of this problem from outside medicine. Fabricated references in academic papers - citations to work that does not exist - have risen sharply as AI writing tools have spread: from roughly one paper in 2,828 in 2023 to one in 277 in early 2026, according to analysis reported by Forbes in May 2026. The instructive detail is not the rate. It is that the fake references were correctly formatted, attributed to real researchers, and carried plausible dates. They did not look wrong. Nothing about them signalled “check me”.
That is about scholarly citations rather than health answers, and I am not going to pretend it measures anything about clinical accuracy. But the failure mode is the one that matters here: the output that is wrong is indistinguishable, on its face, from the output that is right.
Which leads to the only test I would apply. Can you get from the answer to the source in one step? Not “does it cite medical literature” - useful, but that tells you the general claim is sound, not that it read your document correctly. The question is whether it will show you the actual page and the actual line in your actual pathology report, so you can see for yourself what it based the answer on.
If it can, you can verify anything that matters and the AI becomes a way of navigating your own records. If it cannot, you are trusting fluency, and you should treat everything it says as a prompt to go and check rather than as a finding.
Three questions to ask any of them
- Where does my data go, and under whose law? Overseas processing is not automatically bad, but you should know rather than assume.
- Is it used to train models? Every serious product in this space now says no for connected health data. Check that the answer covers what you are actually uploading, and check the setting yourself.
- Can I check an answer against my own document? The one above. If a product cannot do this, it is a chatbot with your records in it rather than a record you can interrogate.
Where exora fits
exora works in Australia, on iOS, Android and web, and it is free. You upload your documents, an AI pipeline reads them and extracts the clinical content into a structured record, and you can then ask questions of your own history in plain language.
The design answer to the verification problem is the part I would judge us on. Every fact in your record links back to the exact page and text of the document it came from, and answers in the chat cite the document they drew on. That is deliberate: we assume the AI will sometimes be wrong, and built on the premise that checking has to be one tap rather than a project. Data is stored in Sydney and stays under Australian jurisdiction, and health data is never used to train AI models - the detail is on the security page.
Where it does not help, plainly:
- We did not build the underlying models. exora uses established third-party AI models, as everything in this article does. What is ours is the pipeline, the structuring and the grounding - not the model.
- The AI still makes mistakes. Source links exist because errors happen, not as decoration. Treat an answer about something that matters as a starting point and check it.
- No My Health Record connection yet. We are working with Services Australia towards one. Today you download your records and upload them, which works and is a manual step.
- It cannot read documents you have not given it. You have a legal right to your records from any provider who has treated you, and this explains how to ask.
- It is not medical advice. It organises your information and helps you understand it. It does not diagnose, it does not recommend treatment, and it is not a substitute for talking to a clinician.
A sensible way to use any of this
Use it to prepare, not to decide. These tools are genuinely good at turning clinical shorthand into English, at reminding you what happened when, and at helping you arrive at an appointment with a clear question instead of a vague worry. That is a real improvement on the status quo, where most people walk in having forgotten two thirds of their own history.
What none of them should be is the last word. If an answer suggests something has changed, or something looks wrong, the next step is a clinician - and ideally a clinician you can show the actual document to.
Product availability, pricing and features were checked on 30 July 2026 and are sourced inline. This category is changing very quickly - ChatGPT Health became broadly available the same week this was written - so treat the dates as load-bearing and check anything important against the vendor’s own current documentation. This article is general information about software, not medical advice. It does not describe or recommend any treatment. Talk to a qualified health professional about your health.
