Every workout,
decoded.
Chiron reads every session from whatever you train with, re-plans tomorrow, and handles the legwork — race entries, fueling plans, benchmarking — in the messaging app you already live in.
Reads every session, re-plans tomorrow, runs the errands — in the app you already text in. MORE +LESS −
Three moves.
No new app.
01 — HOW IT WORKS
Link whatever records your training — watch, head unit or platform. History imports; the model calibrates to you overnight.
Ride, run, swim as planned. Nothing to log, nothing to tag — the data arrives on its own.
Minutes after you sync: what happened, what it means, and what changes tomorrow.
A coach,
not a dashboard.
02 — CAPABILITIES
Say it like you'd say it to a coach.
Send a voice note from the trainer, the trailhead, the pool deck. Chiron hears "legs felt hollow on rep three" and folds it into the model — subjective feel weighed against power, heart rate and recovery, not filed away in a notes field.
Send a voice note. "Legs felt hollow" gets weighed against power, HR and recovery. MORE +LESS −
One link. Refit to every athlete who opens it.
Chiron builds each session from your own data — current threshold, fatigue, the block you're in. Share it with a teammate and the structure holds while every target re-computes to their physiology. Same intent, different watts.
Built from your data. Share the link — every target re-fits to whoever opens it. MORE +LESS −
Thirty seconds.
Zero friction.
03 — ONBOARDING
The science does
the coaching.
04 — METHOD
Stress in.
Adaptation out.
Every session is scored for training stress, then fed into an impulse-response model: fitness builds slowly over about six weeks, fatigue arrives fast and clears in days. The gap between the curves is form — the number that decides whether tomorrow is intervals or rest.
Stress in; fitness builds slow, fatigue clears fast. The gap is form — it decides tomorrow. MORE +LESS −
Chiron re-fits the decay constants to your own history — how fast you actually recover, not how fast the average athlete does.
A watt is not
just a watt.
Performance data only means something against its conditions. Chiron reads the metadata around every file — location and elevation, heat, wind, surface, the timezones you cross, last night's sleep — and normalizes before it judges.
Elevation, heat, wind, surface, timezone, sleep — normalized before it judges. MORE +LESS −
Then it plans forward with the same signals: tomorrow's weather shapes tomorrow's session.
Intensity lives at the poles. Chiron polices the distribution — the most common self-coaching failure is drifting into the dead middle.
Three weeks build, one week absorbs. The ramp is sized by your fitness slope and HRV trend — not by the calendar.
Load falls, fitness holds, form peaks on the start line — timed from your own decay constants, not a generic two-week rule.
His team ear-pricks lactate so threshold work never drifts into racing. That discipline is the transferable part — not the 30-hour weeks. Chiron holds your 2×15′ under the same ceiling, estimated from power–HR drift instead of blood.
Threshold clamped like his team does with lactate — via power–HR drift. +−
The 220km weeks don't transfer — the split does: ~85% genuinely conversational, hard days few and specific. On a 40km week that's the line between improving and being permanently tired. Chiron keeps both sides honest.
The split transfers, not the 220km: ~85% easy, hard days specific. +−
A staff of physiologists keeps his base in Zone 2 and paces racing off his power-duration curve. Both scale down cleanly: your Z2 ceiling and your curve are already in your data — Chiron re-fits them after every ride and sizes efforts to match.
Z2 base + power-duration pacing — your curve re-fits after every ride. +−
Nobody needs her 90km weeks. What transfers: threshold repeats held to the second per 100, and stroke count as an early fatigue tell. Chiron reads both from any pool swim and re-tests your CSS on schedule.
Threshold repeats to the second + stroke count as a fatigue tell. +−
Ask once.
It's handled.
05 — THE AGENT
Agentic AI, not a chatbot: it triggers itself, uses tools, writes to your systems and follows through. Six workflows, live:
Train on evidence.
Chiron is in invite-only beta. Leave an email; we onboard a new cohort every week, in order.