MultiSport Research - The Power of Insight
Home All news Training Tilt links coaching platform with external AI tools

Training Tilt links coaching platform with external AI tools

Training Tilt MCP Server Release

Training Tilt, an all-in-one coaching platform for endurance sports coaches, has confirmed the launch of its Model Context Protocol (MCP) server. The new feature provides a direct connection between the Training Tilt platform and external AI assistants such as Claude and ChatGPT.

According to Training Tilt, the new integration allows coaches to maintain full control over how they incorporate artificial intelligence into their coaching workflows. By granting their preferred AI tools access to coaching data, coaches can manage administrative tasks independently from direct athlete interaction.

Through the new connection, coaches can prompt their own AI to review an athlete’s training load, identify insights or anomalies, build structured interval sessions, or adjust a training calendar. Workouts can also be pushed directly to devices from Garmin, Wahoo, and Zwift via a chat window, eliminating the need to copy and paste information between different software applications.

Additionally, coaches can utilise features built into Claude and ChatGPT, such as projects, skills, and scheduled tasks, to train their AI models and automate manual coaching workflows.

The coaching platform adds that its latest launch marks a distinct approach in a market where several training platforms are developing proprietary AI algorithms to generate and adapt training plans automatically. According to Training Tilt it is choosing not to place its own AI between coaches and athletes. Instead, the MCP server allows coaches to integrate the external AI tools they already utilise and prompt with their own coaching philosophy, ensuring all decisions remain subject to coach approval.

Cam Langsford, Founder of Training Tilt, said “The coaches this is built for are already using Claude or ChatGPT every day. They’ve shaped it around how they coach. What they don’t want is someone else’s algorithm inside their coaching platform, making calls on their athletes in ways they can’t see or control. They want their AI, connected to their data, doing what they tell it. So that’s what we built.”

The new integration is built on the Model Context Protocol, an open standard introduced by Anthropic in 2024 to allow AI assistants to interact directly with external platforms. Training Tilt is among the first endurance coaching platforms to deploy a production MCP server.

The MCP server provides an AI assistant with the same access levels as the coach’s personal login, including athlete profiles, workout libraries, and calendars. While data reading is immediate, any modifications, such as scheduling a session or messaging an athlete, must be displayed to the coach for explicit approval before execution.

Key capabilities available at launch include:

  • Reviewing athlete training summaries, load trends, and planned-versus-completed sessions across any date range to discover new insights
  • Checking across all athlete conversations to discover anything the coach has missed
  • Building structured workouts, including intervals, targets, and recoveries, from a plain-language description or voice input
  • Scheduling, moving, and removing sessions on athlete calendars via text or voice
  • Searching the workout library and saving new sessions as reusable templates
  • Sending planned workouts to athlete devices across Garmin, Wahoo, Suunto, Zwift, and Coros
  • Drafting athlete notes and workout comments for coach approval before sending
  • Using Claude or ChatGPT scheduled jobs to automatically perform tasks on a given schedule, reducing manual data collation

Beyond daily coaching tasks, the platform notes that the MCP server allows coaches to create customised solutions using their training data. Applications observed during testing include:

  • Creating skills and projects to instruct the AI on specific coaching philosophies and techniques, allowing the system to apply those skills to tasks automatically
  • Generating leaderboards from athlete data to maintain social engagement among athletes
  • Creating review images or animations for athletes covering a specific month or year
  • Developing automated workflows using projects and skills to handle administrative tasks while the coach focuses on athlete interaction

The MCP server is currently live for coaches using the platform. The setup process takes approximately five minutes, requiring coaches to add a connection link to Claude or ChatGPT and log in using their Training Tilt credentials. A setup guide and a library of 50 coach-ready prompts are available in the company’s support centre.

Cam Langsford at Training Tilt added “There’s a version of AI in coaching where the platform’s algorithm writes the plans and the coach becomes a spectator. We think that’s backwards. Training Tilt is built for coaches. The AI works for the coach, applies the coach’s philosophy, and changes nothing without the coach’s sign-off. We don’t compete with coaches for their athletes, and we won’t compete with them for their judgment either.”

www.trainingtilt.com