How Mike Norton used AI to build P4 Health’s dashboard in 1 year, not 10

An extraordinary amount of health information is now captured by patient wearables like watches and rings, which measure sleep quality, heart rate variability, blood pressure, exercise habits and much more. It’s valuable data but is largely unused as the healthcare system still only has sight of formal tests like pathology and imaging. 

That fragmentation proved frustrating to Mike Norton when he went to see his doctor about a few minor but ongoing concerns. Rather than reactive healthcare, he wanted a system that connected data points, personalised recommendations and enabled him to take a proactive, preventive approach to his health. 

Norton has now turned that idea into the P4Health platform, using AI to accelerate prototyping, build the P4 Dashboard and complete development. He estimates a traditional, manually coded build would have taken 10 years. Using AI, it took him just 1 year, though he admits there were a lot of late nights in those 12 months! 

Building a platform around disconnected health data

P4 Health brings together multiple forms of personal health data into one platform that informs and educates users so they can take proactive steps to maintain or improve their health. 

As Norton explains, those 4 Ps stand for “personalised, predictive, proactive, and participatory.”

Data from wearables sits alongside biomarker reports, epigenetic insights and microbiome testing that users can order through the platform.

I’ve got epigenetics out of the States. I’ve got dry blood spot tests out of Sweden, and I’ve got a microbiome test kit out of Denmark,” says Norton. “I’ve got over 33 wearables brands that I can pull into this. Garmin, Polar, Dexcom, Oura, Whoop bands — they all get pulled through the machine learning logarithms that I’ve built to give a personalised output.”

But, as he explains, “The biggest thing I had with this, and I struggled with this for so long, because I’ve got so much data, how do I take nearly 3,500 data points and put it on a dashboard?”

The P4 Health dashboard is an elegant solution that organises, interprets and presents health data clearly enough to enable action. Rather than showing users a wall of disconnected metrics, it filters information through specific journeys and views, surfacing the most relevant information according to each context, such as longevity, gut health or menopause. Norton also built in cohort analysis, using comparative data organised by factors such as age, gender and health journey.

“It gives me so much more information than, dare I say, my GP,” says Norton, before quickly adding, “I didn’t say that out loud!”

A founder with a systems mindset

Norton’s career has been anything but linear. Originally a plant operator in the Royal Engineers, Norton has been a DJ, an entrepreneur, a coffee roaster and a designer. In Australia, he worked in Microsoft-certified IT roles, managed infrastructure and support teams, and contributed to large technology rollouts in Brisbane.

The thread there is that I’ve always, always been into IT,” comments Norton, “Even DJing, it’s about matching beats and making sense of logarithms, right?”

That systems mindset runs through P4 Health. It is the work of someone comfortable with complexity, infrastructure and integration, and able to bring multiple moving parts together into a platform that feels coherent and usable. 

Where AI sped up the build

Norton says that without AI, the platform may have taken him 10 years to build manually. Instead, he got it live in roughly 1 year, using AI to help prototype, code and iterate at a speed that would have been difficult to achieve otherwise.

While AI made the build faster, Norton still needed to shape the architecture, decide how data would flow through the system and refine the outputs so they remained useful and commercially relevant. 

The platform itself has several layers. One uses Norton’s own machine learning logic to interpret incoming data and generate educational outputs. Another is Henry, the customer-facing AI that helps ensure a smooth user experience. 

In the dashboard, if you go into health mode, it leans into my local AI, which I’ve called Henry,” says Norton. “He’ll tell you about the dashboard and guide you about next steps, like suggesting you visit the digital twin lab. You can also ask him questions.”

Behind Henry sits a more advanced layer that draws on structured source material Norton has assembled from detailed textbooks and PDFs across multiple categories. He converted that material into machine-readable JSON, stored it within his Cloudflare environment and made it retrievable through his AI worker, allowing richer and more context-aware responses.

Using AI to ease the workload and support member engagement

Some of the most practical lessons in Norton’s story sit behind the scenes rather than on the front end.

He has built two internal AI agents, Bert and Bob, with distinct roles. Bert handles more of the operational side, monitoring the platform, checking whether Henry is active and keeping the AI layer ready so users do not face delays when logging in after a quiet period. 

Bob, meanwhile, aids marketing and member engagement. Bob reviews Norton’s emails, studies member behaviour and drafts personalised weekly emails based on what each user has done on the platform.

As Norton explains: “Bob is my marketer. He reviews my emails and reads research updates that I receive by email – there’s probably a few years of them, all in subfolders about wearable devices, academic studies, marketing, competitors and so on. So, Bob goes and researches my emails and checks on the members in my platform in MemberStack. He looks at the journeys and the interaction that my individual members have had. And then he uses their unique statistics and metadata to create a weekly personalised email based on their context, on their journey. But he can’t send anything out until I’ve reviewed it.”

That kind of application is likely to resonate with healthcare leaders looking for realistic, low-friction uses of AI. Many businesses already have rich internal sources of insight sitting in inboxes, folders, CRM notes and engagement data. Organising that material, extracting useful patterns and drafting more personalised communication is an achievable use case, especially when a human remains responsible for final review and approval.

Norton also experimented with enabling Bob and Bert to communicate with each other, hinting at how AI can start supporting coordination and workflow for busy executives. 

Guardrails, trust and the commercial question

Norton is careful about the limits of the system. P4 Health is intended to be educational and guidance-based, not diagnostic or therapeutic. That distinction is built into the platform and supported by guardrails around how the system operates. His internal bots are also contained within a separate Linux environment safely closed off from his C-drive.

Lessons for healthcare leaders

Mike Norton’s story points to a broader shift in healthcare, where patients are generating increasing amounts of data outside traditional clinical settings and businesses are beginning to ask how that information can be used more meaningfully.

That creates opportunities across the sector:  

• A general practice could use a platform like P4 Health to support more informed conversations with engaged patients.
• A preventive health clinic could use it to keep patients connected between visits and give them a clearer sense of progress over time. 
• A corporate wellness provider could build a more tailored member experience around wearables, testing and education. 
• Established software businesses may also see value in integrating these kinds of capabilities into their own systems rather than building from the ground up.

Norton’s experience also shows how much AI has changed the pace of innovation. Founders can now prototype faster, test ideas sooner and build tools that would once have required much larger teams, longer timelines and deeper funding. That opens the door to more experimentation, though it does not remove the need for commercial judgement. Adoption, product-market fit, trust and strategic focus still decide what happens after launch.

Not every business needs to build a platform as ambitious as P4 Health to benefit from AI. The more immediate opportunity may be using it to organise internal knowledge, personalise communication, surface customer insights, monitor systems and reduce the administrative load that slows teams down.