Pulsai is a web-based tool that converts a user's Garmin Connect data into a secure, hosted Model Context Protocol (MCP) server. This server provides a private HTTP/Server-Sent Events (SSE) endpoint that enables AI applications such as Claude, ChatGPT, and Gemini to directly access and query the user's Garmin metrics. The service eliminates the need for custom pipelines, local setup, or self-managed servers by handling all infrastructure on the user's behalf. The primary features of Pulsai include a hosted MCP endpoint that is unique to each user, encrypted token storage using Laravel Crypt, and a sign-in process that requires only a Google account and a one-time Garmin account link. No Garmin passwords are stored, and users can unlink their accounts at any time. The service supports queries about activities, sleep data, heart rate variability (HRV), and other metrics available through Garmin Connect. Typical use cases involve integrating personal fitness and health data into conversational AI workflows. For example, a user can ask Claude or ChatGPT to summarize recent workouts, analyze sleep trends, or correlate HRV with activity intensity. The workflow begins by signing in with Google, linking the Garmin account, and then copying the provided MCP URL into the MCP client settings of the chosen AI application. Once configured, the AI can interact with the Garmin data through natural language. Technical requirements are minimal: a Google account for authentication, an active Garmin Connect account, and access to an AI client that supports MCP servers. The Pulsai service itself runs as a hosted endpoint, so no local software installation is needed. The service ensures data privacy by encrypting all Garmin tokens at rest and only decrypting them to serve the user's own endpoint.