Murph is an open-source, agentic personal health AI designed to eliminate the friction of modern health tracking by meeting users directly where they already communicate, such as iMessage, Telegram, and email. Rather than requiring users to learn another standalone dashboard or navigate fragmented apps, Murph integrates across popular wearables like Oura, WHOOP, Apple Health, Garmin, and Fitbit, as well as lab bloodwork and meal snapshots. Built on Codex with its own headless browser and telephony capabilities, Murph acts as a proactive executive assistant that can book appointments, call clinics to handle hold times, order supplements, and manage calendar events. Beyond basic logging, Murph differentiates itself through n-of-1 self-experimentation and collaborative accountability. It structures scientific, baseline-driven protocols across sleep, recovery, supplementation, and exercise, evaluating whether interventions actually alter user biomarkers over time. Socially, it acts as a neutral referee in group chats, running personalized challenges based on relative adherence rather than raw stats, and generating family newsletters. Released under the Apache 2.0 license, Murph ensures full transparency, allowing users to self-host their health stack locally or use the managed cloud service with strict encryption standards that do not use personal health records to train general foundation models.

Pros of Murph

Frictionless interaction model operating entirely inside existing channels like iMessage, Telegram, and email without demanding another app install
Agentic real-world capabilities powered by browser automation and telephony to schedule doctor appointments, call clinics, and reorder supplements
Evidence-based, baseline-driven self-experimentation framework that correlates interventions like supplementation or sleep protocols directly with wearable biomarkers
Open-source Apache 2.0 architecture offering self-hosting, local model inference options, and strict privacy boundaries that prevent data from training foundation models
Well-designed group dynamics that referee friendly health challenges based on personal baselines and improvements rather than raw biometric comparisons

Cons of Murph

Autonomous agent actions like scheduling appointments and placing calls may occasionally require user intervention or verification when dealing with complex clinic receptionists
Self-experimentation insights rely heavily on third-party wearable accuracy and consistent user logging, which can introduce statistical noise or false correlations
Advanced self-hosting, custom subagent configurations, and local LLM runtime management require developer familiarity with terminal tools and CLI workflows

Frequently Asked Questions

Do I need to wear a smart watch or fitness tracker to use Murph?

No. While connecting wearables like Oura, WHOOP, Apple Health, or Garmin enriches your data stream, Murph can also operate effectively from lab results, photo-based meal logs, symptoms, notes, and manual workout logs.

Is Murph open source, and can I host it myself?

Yes. Murph is licensed under Apache 2.0. You can inspect the codebase, run the entire stack locally via its CLI on your own computer to keep data strictly on-device, or choose the fully managed hosted service.

How does Murph handle phone calls and booking appointments?

Murph runs autonomous agent tools with browser access and telephony features. It can dial clinic phone numbers, navigate phone tree menus, wait on hold with receptionists, and locate convenient calendar slots, requiring your explicit approval before purchasing or finalizing bookings.

Does Murph use my personal health metrics to train AI models?

No. Murph encrypts health data in transit and at rest, and health information managed on the platform is not sold or used to train general-purpose foundation AI models.

Does Murph provide formal medical diagnoses or prescriptions?

No. Murph provides educational tools, self-experimentation protocols, and organized summaries to assist your personal wellness journey and clinical conversations, but it is not a licensed medical provider or a substitute for professional medical care.