Function, a health data platform based in Austin, Texas, announced on March 20, 2026, that it will expand its AI connector portfolio by integrating with Perplexity Health, allowing members to securely link electronic health records and wearable device data in a single location. The product arrives as the U.S. Centers for Medicare and Medicaid Services has been pushing a national initiative to let patients share medical records across apps, giving the startup’s approach a federal tailwind. But the convergence of clinical records and consumer fitness data in one AI-powered dashboard raises pointed questions about privacy, data security, and whether the technology can deliver on its promise without exposing patients to new risks.
What Function and Perplexity Health Are Building
The core idea is straightforward: patients currently keep their lab results in one portal, their Apple Watch or Fitbit data in another, and their prescription history in yet a third system. Function’s new connector aims to pull those streams together so that medical records sit in one place alongside real-time biometric feeds from wearables. Perplexity Health, the AI layer in this arrangement, is tasked with interpreting the combined dataset and surfacing personalized health insights for the user.
The partnership positions Function as the plumbing and Perplexity Health as the intelligence engine. Function already maintains a portfolio of AI connectors, and the Perplexity integration is the latest addition. For users, the practical change is that a single app could theoretically show a doctor’s notes next to overnight heart-rate trends, flagging patterns that neither data source would reveal on its own. That kind of cross-referencing has long been a goal of digital health advocates, but few consumer products have attempted it at scale with an AI interpretation layer attached.
In Function’s vision, a member might log in to see recent blood test results, medication adherence, and step counts organized along a shared timeline. Perplexity Health would then generate summaries, such as noting how changes in activity levels correlate with blood pressure readings, or highlighting when sleep disruptions precede migraine episodes. The appeal is convenience and context: instead of scattered portals and raw numbers, users receive narrative explanations and suggested questions to raise with their clinicians.
Federal Policy Is Clearing the Path
Perplexity Health’s approach does not exist in a regulatory vacuum. The Centers for Medicare and Medicaid Services has issued guidance on a national initiative designed to make it easier for patients to share medical records across apps and programs. The agency’s interoperability rules require insurers and health plans to offer standardized application programming interfaces so that third-party apps can request and receive patient data with the patient’s consent.
That policy context matters because it effectively lowers the technical barrier for companies like Function. Before these rules, connecting to a hospital’s electronic health record system often required custom integrations that only large health IT vendors could afford. Now, standardized data-sharing protocols mean a startup with the right certifications can plug into the same pipelines. The broader Medicaid program, described on the main federal Medicaid site, has also become a focal point for expanding digital access and modernizing how low-income beneficiaries interact with their coverage.
CMS has expanded resources for people enrolled in public programs, including detailed beneficiary tools that explain how individuals can obtain and share their health data. Participants in the Children’s Health Insurance Program are likewise being encouraged to understand their rights around digital access to records, signaling that the agency expects consumer-facing health apps to become more common, not less.
For people enrolled in plans through the federal marketplace, definitions of qualifying health coverage already anticipate digital tools that help patients manage their care and track benefits. And CMS trains enrollment assisters through a dedicated portal that increasingly covers how beneficiaries can exercise their data access rights when using third-party apps. The regulatory direction is clear: Washington wants patients to control their own records, and it is building the infrastructure to make that happen.
The Privacy Tension No One Has Solved
Yet the same federal push that enables products like Perplexity Health also exposes a gap. Medical records held by hospitals and insurers fall under the Health Insurance Portability and Accountability Act, known as HIPAA. Once a patient voluntarily shares that data with a consumer app, however, the protections can change dramatically. Many consumer health apps operate outside HIPAA’s enforcement perimeter, meaning the data may be governed instead by a company’s own privacy policy and the Federal Trade Commission’s general consumer protection authority.
This distinction is not academic. Wearable data, including sleep patterns, heart rate variability, and exercise logs, can reveal sensitive information about a person’s mental health, substance use, or chronic conditions. Combining that with clinical records creates a richer profile than either dataset alone. If a breach occurs, the exposure is correspondingly larger. And if the AI layer draws inferences, such as flagging a user as potentially pre-diabetic based on glucose trends and family history in their medical record, the question of who sees that inference and how it might affect insurance eligibility or employment becomes urgent.
Function’s announcement states that members will be able to “securely” link their records, but the company has not publicly detailed its encryption standards, data retention policies, or whether it has undergone independent security audits. Perplexity Health’s own technical specifications for handling protected health information were not included in the announcement. Until those details surface, the security promise remains a marketing claim rather than a verified safeguard.
There is also the issue of secondary uses. Even if a platform does not sell identifiable data, aggregated or de-identified datasets can still be valuable to advertisers, pharmaceutical companies, or data brokers. Without clear restrictions, patients may find their health patterns informing marketing campaigns or risk models they never agreed to support. Transparency dashboards, granular consent controls, and easy data deletion tools are all potential counterweights, but they are not yet standard across the digital health ecosystem.
Why Most Prior Attempts Stalled
The idea of unifying health records and wearable data is not new. Apple Health Records, launched several years ago, lets iPhone users download clinical data from participating hospitals. Google attempted a similar play with its now-shuttered Google Health platform before pivoting to enterprise health IT through partnerships with health systems. Neither product achieved the kind of broad, AI-driven synthesis that Function and Perplexity Health are describing.
The reasons for those earlier struggles are instructive. Hospital IT departments have historically been slow to open their systems, partly out of legitimate security concerns and partly because data silos protect their competitive position. Patients, meanwhile, have shown mixed enthusiasm for managing their own records. Downloading a lab result is one thing; interpreting it alongside wearable data requires a level of health literacy that many people lack. An AI assistant that translates clinical jargon into plain language could close that gap, but only if the translations are accurate and the system avoids generating false alarms that send anxious users to emergency rooms unnecessarily.
Trust has been another sticking point. Technology companies entering health care often underestimate how wary patients are about corporate access to their most intimate information. Several high-profile data breaches and opaque partnerships between hospitals and tech firms have made people more cautious about clicking “allow” on data-sharing prompts. For Function and Perplexity Health, overcoming that skepticism will likely require verifiable security practices, clear explanations of how data is used, and visible options to opt out.
Even if the technical and trust hurdles are cleared, sustained engagement is not guaranteed. Many health apps see a spike in usage when users first sign up, followed by a steep drop-off once the novelty wears off. To avoid that pattern, an AI-driven dashboard has to deliver value that feels immediate and personal, catching medication conflicts, spotting early warning signs of complications, or simplifying conversations with clinicians. Otherwise, the connector risks becoming just another unused icon on a crowded smartphone screen.
Promise, Risk, and What Comes Next
Function’s integration with Perplexity Health illustrates both the promise and the unresolved risk of the next wave of digital health tools. On one hand, the technical groundwork laid by federal interoperability rules makes it more feasible than ever for patients to assemble a complete picture of their health. On the other, the shift of sensitive information into consumer apps exposes it to a regulatory patchwork that was not designed for AI-driven analytics and continuous data streams.
Whether this model succeeds will depend less on clever algorithms than on governance. Patients will need to understand what they are consenting to, regulators will need to clarify how existing laws apply to hybrid clinical-and-consumer platforms, and companies like Function will need to demonstrate that their security claims are more than aspirational. If those pieces come together, unified records and wearables could help people and their clinicians make better decisions. If they do not, the latest generation of health data platforms may end up repeating the trajectory of earlier efforts, ambitious in scope, impressive in demos, and ultimately limited by trust they could not fully earn.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
