Experiment in progress

Can AI coach a human to finish a triathlon?

One AI. One untrained athlete. One 100km triathlon. No human coach. No shortcuts. Everything documented.

304
Days left
100
Kilometers
3
Disciplines
0
Human coaches
The Premise
What happens when you hand your body to an algorithm and say: get me to the finish line?
02T is a public experiment. I'm using AI as my sole endurance coach to prepare for the T100 Qatar Triathlon World Championship Final. Not a fitness influencer. Not a pro athlete. A university senior with a startup and zero triathlon experience.
The AI builds my training plan, adjusts daily sessions based on Oura Ring recovery data and Strava metrics, runs a rules engine to prevent overtraining, and generates post-session coaching notes. Every decision is logged. Every adaptation is transparent. If AI can do this, it changes how millions of people access coaching.
The Race
T100 Qatar World Championship
The same distance the world's best pros race. 2km swim in the Arabian Gulf, 80km bike through the streets of Lusail, 18km run past the iconic Lusail Stadium. December in Doha. 30+°C. Full closed roads.
Triathlon World Tour
Qatar T100 Triathlon
11-12 December 2026 · Lusail, Doha, Qatar
🏊
2km
Arabian Gulf
🚴
80km
Lusail streets
🏃
18km
Lusail Boulevard
The Athlete
Ekaterina
Not a triathlete. Not yet. CEO of a restaurant tech startup, senior at Carnegie Mellon, building an AI health company on the side. Background in shooting sports (regional champion, St. Petersburg), equestrian training at Al Shaqab. Zero endurance sport background. That's the point.
Based in
Doha, Qatar
Age
21
Endurance background
None
Sport background
Shooting, Equestrian
Currently building
SUFRA (AI restaurant ops) · Basira Health Intelligence (GLOW)
The Format
How this works
Everything is public. The training, the data, the struggles, the AI decisions. 02T is both an experiment and a series.
01

AI builds the plan

44-week periodized training plan across five phases. Base, Build, Peak, Taper, Race. Every session has a purpose, HR cap, zone target, and RPE ceiling.

02

Body data drives decisions

Oura Ring pulls readiness, HRV, sleep, and resting heart rate daily. Strava captures every session. A rules engine evaluates whether to push, maintain, or pull back. No guessing.

03

Assess every 3 weeks

Run pace at fixed HR. Bike HR at fixed effort. Swim 400m time trial. Three tests that prove (or disprove) whether the AI's approach is working.

04

All of it, documented

Weekly content drops. Raw training data, coach notes from the AI, body metrics, honest reflections. The 02T series follows the full journey from zero to race day.

The Rules
What makes it real
R1 No human coach. All programming, adjustments, and decisions made by AI.
R2 Full data transparency. Every metric, every session, every decision is published.
R3 Safety guardrails are non-negotiable. The AI has hard rules for injury prevention and overtraining.
R4 The goal is to finish. Not to win. Cross the line within cutoff. That's success.
R5 If it doesn't work, that's a valid result. Failure is documented the same as progress.
304 days until the AI's coaching is tested at the finish line

Follow the experiment

Weekly drops. Raw data. No fluff. Watch an AI try to do what only human coaches have done.

📱 Instagram ▶ YouTube 𝕏 X / Twitter