Athlete Space
Athlete Space is an AI coach for endurance athletes. I founded it in 2025 and built it myself. It reads an athlete's training data, builds their plan and adjusts it as they train.
Visit athletespace.ai
Athletes have plenty of data and little guidance
Endurance athletes record heart rate, power, pace, sleep and years of training history. Most of them still follow a static plan, or see a coach once a week.
A coach built on training science
A deterministic pipeline computes training load and builds the plan. A set of LLM agents explains it, answers questions and adapts it. The agents work from the athlete's own numbers, and every reply is verified before it is sent.
How it is built
Planning pipeline
Seven stages, from season plan to daily calendar. Each stage makes one LLM call and has a deterministic fallback.
Coaching agents
Separate agents plan, advise, evaluate and verify. Each has typed, structured outputs and its own evaluations.
Release process
Changes ship behind feature flags, against a regression suite, with a tested rollback plan.
Why I built it
I have trained seriously for a decade. I ran 2:29 at the St. Jude Memphis Marathon in 2025, and finished fourth overall and first in my age group at Ironman Maryland in 2022, in 9:01, which qualified me for the World Championship in Kona. I built Athlete Space because I wanted this coach for my own training.