Appendix Lets Patients Get Prescriptions from Claude Without Ever Speaking to a Doctor

Appendix lets users have an AI agent draft a medical encounter, then routes it to a human physician for review and a prescription if warranted.

By · Published

Why it matters

Appendix is an early example of AI agents being pointed at regulated healthcare workflows, where the commercial opportunity depends on keeping a licensed human in the loop.

Appendix Lets Patients Get Prescriptions from Claude Without Ever Speaking to a Doctor — Appendix lets users have an AI agent draft a medical encounter, then routes it to a human physician for review and a prescription if warranted.

Benjamin Jack, MD (@bjackmd), an emergency room physician, said in a three-post thread on X that he has built Appendix, a service meant to turn Claude Code or another AI agent into the front end for a prescription visit.

https://x.com/bjackmd/status/2062200039775023489

Jack described the workflow as an intake system, not an autonomous prescribing bot: a patient tells an AI agent what is going on, the agent writes up the encounter and submits it through Appendix's API, and a board-certified physician reviews it. If the doctor decides a prescription is appropriate, Jack says the prescription goes to the user's pharmacy.

"There's no chatbot - it's a human doctor in-the-loop for your AI agent," Jack wrote. That framing matters because the pitch is not that an AI model can prescribe medicine on its own. Appendix is trying to sit between consumer AI tools and licensed medical review, making the agent responsible for documentation while keeping the clinical decision with a physician.

The business model is partly free, according to Jack. Appendix's medical knowledge base is free, while physician review and prescription fulfillment require a subscription for an individual or family. Jack did not disclose pricing, the states or jurisdictions covered, the physicians in the review network, or how Appendix handles identity, medical history, contraindications and follow-up care.

The sharper question is whether patients will trust an AI agent to draft the visit accurately enough for a clinician to act on it, and whether regulators will view the API layer as documentation software, telehealth infrastructure or something closer to a new prescribing channel.

Reader comments

Conversation for this story loads after sign-in.