A Seattle company backed by Y Combinator says it has solved one of digital health's most stubborn problems by deploying voice and browser automation when electronic data exchange hits a dead end.
Hubble launched an API earlier this year that claims to retrieve patient medical records from more than 70,000 providers, blending traditional electronic health record connections with AI agents that call records departments, navigate patient portals, and submit fax requests. The approach addresses a gap that has bedeviled health tech developers: major EHR vendors and health information exchanges cover large hospital systems reasonably well, but thousands of small practices and specialty clinics remain unreachable by purely digital means.
The company's founders come with pedigree. CEO Prabha Dublish was the first product hire at Grow Therapy, which hit a $3 billion valuation in March 2026 according to Bloomberg, and previously worked at Meta. CTO Aaron Leon spent years at Amazon One Medical, where he led a 50-person engineering team building the company's proprietary EHR and AI infrastructure for over a million patients. He also developed the data platform for Amazon Care before that venture shuttered.
The technical mechanics
After a patient verifies their identity once, Hubble's system fans out across multiple channels. It queries EHR patient portals such as MyChart, pulls data through national health information networks like Carequality and CommonWell, then deploys voice and browser agents for holdouts still operating on fax machines or locked behind analog workflows. The company says every request runs under the patient's individual right of access granted by HIPAA regulation 45 CFR 164.524.
What comes back is a single normalized record with source-level attribution for each data field. Customers can download results in FHIR R4 or PDF format. Hubble also offers an MCP interface, allowing developers to plug agentic workflows directly into the assembled record—a nod to the current wave of companies building decision-support tools on top of patient data.
"Your user verifies their identity once, and we assemble their records… and return them for your AI agent to use," the company explained in its Y Combinator launch post from August.
Early momentum, unanswered questions
Within four weeks of entering Y Combinator, Hubble claims to have signed ten customers and projects seven-figure first-year revenue, according to its company page. The startup has not named any of those customers publicly. Leon stated in August that the company closed what he described as an oversubscribed pre-seed round but declined to share investor names or the amount raised.
Hubble's case studies, which also keep customer identities confidential, claim the platform has cut records retrieval time from months to one week for legal cases and reduced underwriting turnaround to days. Those are meaningful improvements if they hold up at scale, though the absence of named customers makes independent verification difficult.
A crowded, litigious field
The competitive landscape here is both crowded and contentious. Particle Health claims access to records for more than 250 million unique patients through national networks. Human API takes a different tack, aggregating records by automating patient portal logins. Health Gorilla found itself on the receiving end of an Epic lawsuit in January over allegations it improperly accessed roughly 300,000 patient records, according to STAT News. Traditional release-of-information vendors such as Datavant and Ontellus handle manual retrieval workflows that Hubble believes it can automate away.
Redox operates an integration platform with thousands of active EHR connections but focuses on provider-to-provider data exchange rather than patient-mediated retrieval. The distinction matters. Patient-mediated access carries different regulatory protections and obligations, particularly around consent and the individual right-of-access provisions in HIPAA.
The health information exchange landscape remains fragmented despite years of federal investment in interoperability standards. Small practices often lack the resources or technical infrastructure to participate in national networks. Some simply refuse, citing administrative burden or privacy concerns. That leaves a long tail of providers effectively invisible to digital retrieval systems, creating headaches for everyone from insurers running underwriting models to legal teams assembling records for malpractice cases.
Hubble's bet is that AI agents can bridge that gap more efficiently than human coordinators chasing faxes and phone trees. Whether voice and browser automation can scale reliably across tens of thousands of providers with wildly varying systems and workflows remains an open question. So does the matter of cost—the company has not disclosed pricing, and deploying AI agents to make phone calls and navigate portals is unlikely to be free.
The company says it is live in production and offers 20-minute demos tailored to specific use cases. For digital health developers tired of stitching together incomplete patient histories, that might be enough to warrant a closer look.
