N1 for Claude.

What it is

N1 for Claude is a connector built on the Model Context Protocol. Once connected, Claude can read the records you hold in N1 and answer questions about them in the conversation. It does not replace the N1 app. It brings the parts you reach for during a consultation, the history, the latest values, the reports, into the place you are already writing.

The connector works the way N1 works. The model reasons and writes; N1 supplies the numbers and checks them. Charts, tables and reviews are rendered by N1 inside the chat, so what you see is the same as what you would see in the app. Generating a report through Claude uses your N1 quota exactly as generating one in the app does.

What you can ask

Each prompt below can be typed as it is. Claude asks whose record a document belongs to and which units you work in before it shows anything.

“Give me a pre-visit briefing for Patient B.”

The patient’s overview: active diagnoses and medications, the most recent readings with their units, reference ranges and dates, anything out of range, and the latest report if one exists. Each value is dated, because a record is a point in time, not the patient’s current state.

“How has his LDL cholesterol moved over the last two years?”

Every dated reading of that marker, with units and the reference range that applied, and the change between them. Ask in mg/dL or mmol/L and the values are converted for you.

“Here is his lipid panel from today. Extract it with N1.”

Paste the report or attach the file. Claude first asks whose record this is and which units you work in. It then shows the report as a table of biomarkers, diagnoses, medications and procedures, each value checked against the printed text and flagged against the report’s own reference ranges. Nothing is saved to N1 at this step.

“Review it against his N1 history.”

A chart for each biomarker that has history, with N1’s readings and the new document’s value on the same line. Under each chart, a table of the N1 value, the document value and the change. The findings follow the charts. As you share more reports, the charts keep building.

“What reports exist for him, and what did the last one say about kidney function?”

The list of generated reports with their dates and status, and the relevant section of the one you ask about. You can also start a new report from a template and check on it later.

“Add a diagnosis of type 2 diabetes, diagnosed March 2024.”

Claude shows you exactly what will be recorded and asks. Only after you agree is the row added. The same two steps apply to editing or removing a reading, diagnosis, medication or procedure. Claude never records something because a document mentions it; only because you asked.

Limits

• One patient at a time. Name the patient when you change topic.
• Lists come 25 rows at a time. Ask for the next page when you need it.
• A pasted report can be up to about 60,000 characters, roughly twenty printed pages.
• A chart needs at least two dated values of the same test.
• Claude can add, edit and remove clinical rows after your confirmation. It cannot delete a patient or change billing, plans or account settings.
• Extraction and review show and check; they do not file the document in N1. Upload the file in the app when you want it on the record.

Privacy and data

The connector sends Claude only what you ask for in the conversation. That information reaches Anthropic and is governed by their terms and privacy notice. Documents you paste into a chat are processed by Claude, not uploaded to N1. The connector keeps no conversations, documents or patient data, only a short-lived request log with identifiers and counts. The full statement is section 14 of our Privacy Notice.

Healthcare organizations should confirm that their agreement with Anthropic covers protected health information before connecting.

Support

Use our contact page. Tell us the patient’s initials, what you asked, and what Claude answered. Do not include the record itself or any health information.