World Tour teams are quietly using AI to revolutionise how cyclists train, and it's fundamentally changing the vocabulary we use to talk about performance. Paul from Vecta reveals how their platform is replacing outdated metrics like FTP and TSS with smarter, science-backed models that adapt daily—and why the gap between pro teams and amateur riders could widen dramatically if this technology stays exclusive.
Key Takeaways
- Critical power model (CP, W', peak power) replaces single FTP metric, giving accurate performance picture across all effort durations—not just one-hour power
- Adaptive training zones auto-adjust daily based on latest data, eliminating the need for manual FTP retests every 8-12 weeks
- AI automatically detects interval structure in outdoor rides without manual lap markers, then compares against similar sessions to quantify progress over weeks
- Volume and intensity are tracked separately to avoid TSS's flaw: two completely different sessions (criterium vs café ride) getting identical stress scores
- Coaches are the primary users because AI saves them hours on session analysis and comparison—enabling them to scale their coaching business
- The platform simplifies complex sport science into two intuitive metrics (volume and intensity) while keeping sophisticated mathematical models running in the background
Expert Quotes
"Everything relies on FTP. So if you measure it wrong or don't update it, everything is wrong. We're trying to have a metric that isn't linked to any threshold value—just kilojoules and weight—so you can see progress without the threshold component corrupting the data."
"I did a full gas 20-minute effort every Tuesday, and I got very good at full gas 20-minute efforts. I hit 405 watts, created zones off that, and towards the end of the year couldn't get near those numbers in zone 5. They were wildly off."
"We simplify down to the simplest form but not any simpler. You can have 125 CTL but feel better at 80 CTL—the model doesn't validate your actual performance. It's trying to quantify the stimulus differently based on volume and intensity distribution."