freddy Launches the First Health & Performance Context Layer for LLMs

AI assistants can reason about almost anything, except you. They have no idea how you slept, how hard you trained, what your glucose did overnight, or even the air you’re breathing. freddy fixes that. Today freddy announced general availability of the first health context layer for large language models, a single private connection that lets any AI assistant read your own health data.

freddy is not just a wearable connector. It brings together the full picture of a person’s health: rings and watches like Oura, Garmin, WHOOP and Polar; training platforms like Hevy, Strava and Intervals.icu; continuous glucose monitors; smart scales and body composition; and home sensors that track your air quality and sleep environment.

Until now, all of this lived in separate apps that never talked to each other. Someone tracking sleep on Oura, training on Garmin, glucose on a CGM, and air quality at home had no way to bring those signals together, and certainly no way to put them in front of an AI assistant. freddy is the first hosted service to do it, and the first built for individuals rather than developers.

Setup takes a couple of minutes. A user connects their accounts through each provider’s official login, then pastes one personal URL into an MCP-compatible AI assistant such as Claude or ChatGPT. From that point on, the assistant carries that person’s sleep, training, recovery, glucose, body composition, and environment as context in any conversation, with new sources added regularly.

freddy is built on a simple principle: people should own the context about their own bodies. Users grant access to their own accounts through each provider’s official login, and freddy reads that data on their behalf. It does not broker, resell, or pool data across users, and any connection can be revoked at any time and data deleted with one click.

The launch lands as AI assistants and personal health converge. Major platforms have signaled that conversational AI will become a primary interface for health, but most efforts are locked to a single assistant or a single device ecosystem. freddy is source-agnostic and assistant-agnostic by design, a neutral context layer that sits between everything a person measures and the AI they ask.

“An AI assistant can reason about almost anything, except you,” said Tom Tomaszewski, founder of freddy. “It doesn’t know how you slept, how hard you trained, or what your glucose did last night. freddy gives it that context, from your wearables to your training to the air in your bedroom, and soon your bloodwork. Connect your sources once, then talk to your own data the way you’d talk to a coach.”

freddy is available today at freddy.coach.