Building an AI Coaching Platform for Endurance Athletes

Overview

In 2025, DICLODE began developing an AI-powered coaching platform designed to help endurance athletes improve their performance without the cost of hiring a traditional coach.

Many endurance athletes — including runners, cyclists, swimmers, and triathletes — train independently. While they often understand the general structure of effective training, they lack the ongoing guidance, feedback, and accountability that a professional coach provides.

The platform was designed to bridge this gap by providing an AI-powered coaching system capable of generating personalized training plans, adapting those plans based on athlete feedback, and providing ongoing coaching guidance throughout a training cycle.

The platform launched in December 2025 and is currently operating as an early-stage MVP focused on validating the coaching model and user experience.

The Challenge

Self-coached athletes frequently encounter several challenges that limit their long-term progress:

  • difficulty designing effective long-term training plans
  • lack of objective feedback on training progress
  • uncertainty about recovery and training load
  • limited accountability for completing workouts

Professional coaching addresses these challenges, but often costs thousands of dollars per year, putting it out of reach for many athletes whose goals are focused on gradual improvement rather than elite competition.

The goal of the platform was to create an accessible coaching experience that provides the structure, feedback, and accountability of a coach while remaining affordable and scalable.

The Solution

DICLODE built an AI-driven coaching platform that generates structured endurance training plans and adapts those plans based on athlete feedback and performance.

The platform focuses on several core capabilities:

  • generating long-term training plans leading up to race events
  • delivering weekly structured workouts
  • collecting athlete feedback on completed workouts
  • providing ongoing coaching guidance and accountability

Through weekly check-ins and ongoing messaging, the AI coach continuously adapts the training plan to better align with the athlete’s progress and constraints.

Core Platform Features

AI Training Plan Generation

The platform generates personalized endurance training plans that provide both:

  • long-term structure for a full training cycle
  • detailed weekly workouts tailored to the athlete’s goals

Plans adapt over time based on athlete feedback and upcoming events such as races.

Weekly Coaching Check-ins

Each week, athletes complete a check-in with the AI coach.

During this process, the platform:

  • reviews the previous week's workouts
  • evaluates how the athlete felt during training
  • asks about fatigue, recovery, and schedule constraints
  • generates an updated training plan for the next week

This creates an ongoing coaching loop similar to what athletes experience with human coaches.

Workout Feedback and Accountability

Athletes record details about each completed workout, including:

  • perceived effort
  • completion status
  • subjective feedback about the session

The system uses this information to provide accountability and ensure training loads remain sustainable.

Race Preparation Guidance

The platform helps athletes prepare for upcoming races by:

  • structuring training blocks around race dates
  • adjusting intensity and volume during peak phases
  • managing taper periods leading into events

This ensures athletes approach race day with appropriate preparation and recovery.

Technical Architecture

The platform was built to support rapid iteration while enabling experimentation with different AI models.

Core Technologies

  • Node.js backend services
  • Next.js / React frontend
  • PostgreSQL database hosted on Supabase
  • Vercel hosting infrastructure

AI Model Integration

The system is designed to work with multiple AI providers, allowing the platform to switch models based on performance and capability.

Current integrations include:

  • OpenAI models
  • Anthropic models

This architecture allows the platform to evolve alongside improvements in AI capabilities.

Event and Automation Infrastructure

Several supporting services enable asynchronous actions and scheduling.

Key services include:

  • Trigger.dev for background workflows and scheduled tasks
  • Resend for transactional email delivery

These tools allow the system to automate actions such as weekly coaching check-ins and training plan updates.

Key Product Challenge: Effective AI Coaching

One of the most interesting challenges has been designing prompts and workflows that enable AI models to behave like effective endurance coaches.

This requires the system to:

  • incorporate athlete history into coaching decisions
  • interpret subjective athlete feedback
  • balance training stress and recovery
  • provide actionable guidance rather than generic advice

A significant focus of development has been refining how past training data is incorporated into future plan generation so the AI coach can continuously improve its recommendations.

Early Results

The platform is currently operating as an early-stage MVP focused on validating the coaching model.

Early testing includes a small group of athletes using the system as part of their regular training.

Initial outcomes include:

  • 5 active early users
  • reported performance improvements from all participants
  • positive feedback on coaching recommendations and weekly guidance

While the platform has not yet introduced monetization, early users report that the system provides valuable structure and accountability for their training.

Future Development

Several features are planned to further enhance the platform.

Upcoming capabilities include:

  • integrations with training platforms such as Garmin and Strava
  • automated workout data ingestion
  • deeper performance analysis
  • expanded coaching intelligence based on historical data

These improvements will allow the AI coach to better understand athlete workloads and adapt plans more precisely.

Why This Approach Works

The platform combines structured endurance training principles with modern AI capabilities to create a scalable coaching experience.

Key design principles include:

  • focusing on athlete feedback loops
  • structuring coaching interactions around weekly planning cycles
  • enabling flexibility in AI model selection
  • prioritizing rapid iteration during the MVP stage

By focusing on practical coaching workflows rather than generic fitness advice, the platform provides meaningful guidance to athletes pursuing long-term improvement.

Work With DICLODE

DICLODE builds advanced platforms that combine AI, data-driven systems, and scalable infrastructure.

If your organization is exploring:

  • AI-powered coaching or advisory platforms
  • personalized recommendation systems
  • data-driven health or performance applications
  • intelligent automation tools

DICLODE can help design and build the technology to bring those ideas to life.