Software Engineer Builds AI Trainer For Two Dollars Monthly. He Vibe Coded It. The Rest Of You Are Still Jogging Blind.
Joel Richards, a software engineer, used AI to build a personal trainer that analyzes his Garmin and Strava health data to optimize physical performance. He reportedly did this for approximately two dollars a month through what he calls vibe coding.
This demonstrates the concept of personal API orchestration. Garmin and Strava both expose health and activity data through developer interfaces. Anyone with modest coding ambition can pipe that data into a language model and receive individualized training recommendations. The mechanism is compositional leverage: cheap AI inference plus existing wearable data equals a bespoke coach without the boutique price.
Joel Richards, an Australian software engineer who combined Garmin and Strava data with AI to create a personal training system for roughly two dollars monthly.
- Create a free OpenAI or Anthropic account and export your last 30 days of activity data from Strava as a CSV file through Strava's data export feature.
- Open ChatGPT or Claude and paste your exported data with the prompt: 'Analyze my recent training patterns and suggest three specific adjustments to improve performance.'
- Compare the AI's recommendations against what your Garmin or Strava app currently suggests, noting where the personalized analysis differs from the generic app guidance.