Your Patients Can Now Bring Their Chart Into Claude
athenahealth just let patients connect their own records to Anthropic's Claude. Here is what a small practice should get ready for, and the one rule your staff need.
By John Martines · November 4, 2026 · 6 min
On November 2, athenahealth launched a connector that lets patients bring their own health records into Claude. Patients will show up with AI summaries of your notes. Here is how a small practice prepares, in order.
A patient sits down for a follow-up at a family practice. Before the nurse can start, he hands over his phone. On the screen is a neat summary of his last three visits, his lab results explained in plain English, and five questions, numbered. "I asked Claude to go through my chart," he says. "It said to ask you about my A1C."
Nobody at the practice sent his chart anywhere. He did, from his own phone, with his own login.
That scene became possible for a large group of patients this week. If your practice runs on athenahealth, or you see patients who use it elsewhere, it is worth a short staff meeting.
On November 2, 2026, athenahealth, which makes practice software used by small and mid-size medical offices, launched athenaPatient MCP. It is a connector that lets a patient link their own athenahealth health records to Claude, Anthropic's AI assistant.
MCP stands for Model Context Protocol, the standard AI assistants use to connect to other software. Here it works like this. A patient with a paid Claude Pro or Max plan in the US adds the connector in Claude's settings and signs in with their athenahealth patient credentials. They choose which information to make available and can take that access back at any time. Then they can ask questions in plain words, and Claude answers from their actual record: lab results, vitals, medications, conditions, immunizations, visit history, diagnostic reports, allergies, procedures and care plans.
athenahealth says its network serves more than 170,000 clinicians. Not every one of their patients will connect, and the patient has to choose to. But the door is open.
How it got here:
- January 7, 2026: OpenAI launched ChatGPT Health, a separate space in ChatGPT where US users can connect medical records and wellness apps, with records connections run by b.well.
- January 11, 2026: Anthropic launched Claude for Healthcare. US Pro and Max subscribers could connect records through partners, opt in, choose what to share and revoke it, and Anthropic said that health data would not be used to train its models.
- March 2026: athenahealth said it would preview a patient MCP server at the HIMSS health technology conference.
- August 2026: athenahealth cites a Pew Research Center survey from that month that found 34% of US adults have used AI assistants for something health-related.
- November 2, 2026: athenaPatient MCP launched for Claude, which athenahealth calls "just the beginning."
This is the patient's tool, not yours. That shapes everything else.
Patients will arrive better prepared, and sometimes worried. An assistant that reads a chart can explain a lab value well. It can also flag something that isn't urgent as if it were. Your staff will hear "Claude said..." more often, and they need a calm answer ready.
Your notes have a new reader. Patients have had portal access to their records for years. Now they have a tireless explainer sitting next to them. Old problem lists, a medication that was stopped but never removed, shorthand that reads badly out of context: all of it will get explained back to the patient, accurately or not.
The privacy line moves to the patient. US health privacy guidance from HHS says that when a patient directs their own records to an app that your practice didn't provide, your practice generally isn't responsible for what that app does with them afterward. The same guidance says a practice generally can't refuse a patient's request on the grounds that it doesn't like the app. In this case the patient connects through athenahealth and Claude, not through your staff. Your job is not to approve or block it. Your job is to keep your own staff on the right side of the line.
Our guide [AI for Dental and Medical Offices](/articles/ai-for-dental-medical-offices-take-repeat-questions-off-the-front-desk) covers using AI at the front desk. This is the other side: AI the patient brings in.
1. Tell your staff it exists. Five minutes at the next huddle. Patients can now connect their athenahealth record to Claude on their own. Nobody at the practice set it up, and nobody needs to.
2. Give the front desk and nurses one line. When a patient shows an AI summary: "Thanks for bringing that. Let's go over it with the doctor." Staff shouldn't confirm or dispute what the assistant said. The clinician does that in the visit.
3. Write the staff rule, and keep it short. The patient may use any AI tool they like with their own record. Staff may not paste patient information into a personal ChatGPT, Claude or Gemini account, ever. If the practice uses an AI tool, it's one the practice chose and signed a business associate agreement for. Put it in writing and have everyone sign it.
4. Tidy the parts of the chart patients read most. Ask each clinician to spend a few minutes on their active patients' medication lists and problem lists. Remove what's stopped or resolved. It was always good practice. Now it shows up in an explanation the patient reads at 11 at night.
5. Plan for longer question lists. Patients who prepare with an assistant often arrive with five questions instead of one. Some practices already ask patients to send questions ahead through the portal. If you don't, this is a good month to start, so the visit doesn't run over.
Whether you use athenahealth or another system, ask:
- Can our patients connect their records to an AI assistant through your patient portal today, and which assistants?
- What exactly can the assistant see, and can the patient limit it?
- Will we see, anywhere, that a patient has connected an assistant?
- Do clinician notes flow to the patient's assistant as soon as they are signed, or after a delay?
- What do you recommend we tell patients who ask whether it's safe?
athenahealth says this is a first step, and others are moving the same way: OpenAI and Anthropic both built health spaces into their assistants this year. Expect more patients, in more practices, to bring their own records into an assistant before the visit.
That can be a good thing. A patient who understands their own labs asks better questions. The practices that do well with it will be the ones that saw it coming: staff with a calm line ready, a written rule that keeps patient data out of personal AI accounts, and charts that read well to the patient who owns them. Hold the five-minute huddle this week.
---
Sources (checked November 4, 2026). athenaPatient MCP, launch date, Claude Pro and Max in the US, sign-in with athenahealth credentials, data types, choose and revoke access, the Pew figure as cited, and "just the beginning": "athenahealth launches new solution to let patients connect their health records to Anthropic's Claude" (athenahealth via Business Wire, Nov 2, 2026); also reported by Fierce Healthcare and HIT Consultant (Nov 2, 2026). Earlier patient MCP preview: athenahealth via Business Wire (Mar 2026). Network size: "Who we are" (athenahealth, accessed Nov 4, 2026). ChatGPT Health and b.well: "OpenAI launches ChatGPT Health to connect data from health apps, medical records" (Fierce Healthcare, Jan 7, 2026). Claude for Healthcare, opt-in, revoke and no training on health data: "Anthropic launches Claude for Healthcare" (MobiHealthNews, Jan 2026). Provider responsibility once a patient directs records to an app, and the duty to send even when concerned: "The access right, health apps, and APIs" (HHS Office for Civil Rights, FAQ guidance).