Athenahealth opens patient records to Claude through new MCP connector

What Athenahealth launched

The connector uses the Model Context Protocol, an interoperability framework that lets AI systems reach external data sources. Athenahealth says Claude is the first assistant it supports, not the last.

Patients sign in with their existing athenahealth patient portal credentials and pick which categories of medical information they want to share. The company says it never hands those credentials to Claude and encrypts data in transit. Once connected, Claude can reference laboratory results, medications, vital signs, health conditions, diagnostic reports, immunizations, visit history, allergies, procedures and care plans. Athenahealth also says it will limit athenaPatient MCP access to AI assistants that do not train their models on patient information.

Patients keep control, according to Paul Brient, the company's chief product and operations officer. He said they can withdraw permission at any time.

Why patients and rivals are watching

Demand is already there. An August 2026 Pew Research Center survey found that 34% of US adults had used AI assistants for health purposes, including understanding symptoms and diagnoses.

Athenahealth is not alone in pushing records toward consumer AI. OpenAI's ChatGPT for Healthcare has integrations with Epic's EHR that let authorized clinicians access and summarize patient information. Oracle has an AI-powered patient portal that lets users query their records conversationally. In Korea, Hanmi Pharmaceutical said on 2 November that it signed a partnership on 30 October. The partnership moves its Epiday health coaching chatbot off an overseas cloud and onto a Korea-style large language model running on domestic infrastructure. The direction of travel is the same: the record leaves the portal.

The control question

Athenahealth's announcement described access controls and encryption. It gave no accuracy figures for Claude's answers.

Other vendors in the same position are testing before they switch. The four companies have completed testing in a validation environment. They are comparing answer accuracy and response speed between the existing and converted systems, using a standard problem set and an automated scoring framework. Hanmi will finalize timing and scope only after quality and safety verification.

Records are still fragmented

The connector solves a last-mile problem on top of a messy base. PATH's analysis found more than 70 health data systems in one country assessment. PATH prepared the analysis for a Gates Foundation convening in Johannesburg in August. Many of the systems operate independently and cannot exchange information. PATH said "Capture once, use many times" has become a guiding principle for countries trying to cut duplication and improve data quality.

A systematic review in BMC Digital Health covered 35 studies in 18 countries between 2015 and 2025. It found mobile health in 16 of them, DHIS2 in five, electronic health records and AI in four each, and telemedicine in three. The authors put the spread down to health system readiness. Where implementation worked, results were measurable. SMS reminders were associated with 25% to 35% increases in childhood immunization coverage in rural Ghana. DHIS2 improved reporting timeliness by 21% in Ethiopia.

Vendors keep shipping. Qyire launched an electronic health record built for eating disorder services, including automated Medical Emergencies in Eating Disorders scoring.

What happens next

Athenahealth plans to extend the connector beyond Claude to other models. Hanmi will decide how broadly to apply its domestic system once it completes verification.

Elsewhere the money is still mostly pipeline. Healthcare Triangle said on 2 November it had identified an estimated USD 20 million to USD 45 million opportunity in Indonesia over 24 to 36 months. It has signed no definitive agreements. Mulk Technology launched a digital hospital platform with 22 AI-enabled modules and patient support in more than 40 languages.