TRI-HARDER
About

The story behind TRI-HARDER

TRI-HARDER started as a personal experiment: what happens when you give an AI access to your actual training data? Not generic “run more, swim better” advice, but a coach that knows your workouts, your goals, your fitness history, and your constraints.

Most triathlon coaching platforms are either expensive human coaching or rigid plan generators that don't really understand your training. TRI-HARDER sits in between — a conversational AI coach that analyses your real data from Strava or intervals.icu and gives you specific, contextual guidance.

The app is free to use and privacy-first. Your training data stays in your browser. The AI model key lives server-side. There are no ads, no third-party analytics, and no selling of your data.

Built by Martin Chan — triathlete, data scientist, and the kind of person who thinks CTL/ATL charts are genuinely exciting.

Tech stack

FrontendReact (Create React App), Tailwind CSS
AIDeepSeek-V3 via Azure AI Foundry
BackendAzure Functions (Node.js)
HostingAzure Static Web Apps
Data sourcesStrava API, intervals.icu API
StorageAzure Table Storage + localStorage
ReportsjsPDF (client-side PDF generation)

Recently shipped

  • Training Plan page — season phase timeline, phase cards, and milestone tracker
  • Strength & Conditioning guide — Essential 6 exercises with progressions and technique cues
  • Proactive coaching alerts — ACWR overtraining risk, discipline gap nudge, race-week briefing
  • Coaching memory check-ins — inline prompts to confirm or dismiss stale stored facts
  • Free Strava support — training load estimated from HR and power when Relative Effort is unavailable
  • intervals.icu OAuth — connect via OAuth in addition to API key
  • 6 months of Strava history — up from 6 weeks for richer seasonal analysis

What's next

TRI-HARDER is actively developed. Here are some things on the horizon:

  • Scheduled or emailed coaching reports
  • Wearable data integration for recovery and readiness
  • Community features and sharing
  • Custom report templates
  • Enhanced training plan auto-adjustment based on completed workouts

Have an idea? Open an issue on GitHub — contributions and feedback are welcome.