How Reviews Work

The ride review pulls together:

• What the session was supposed to do (target power, intervals, duration, which gap it was addressing)
• What actually happened (your power, heart rate, duration, zone distribution, decoupling)
• If it was an interval session: how each rep compared to target, whether you faded across reps, how you paced within intervals, recovery quality between reps
• Your recent 2 weeks of training — did you complete what was planned? Any patterns?
• Your RPE and any context you shared about the ride
• What's coming next in the plan
• Your coaching notebook — accumulated observations about how you respond to training

The AI generates a short, direct review — 2-3 sentences, in the voice of a human coach. It never references internal system terms. It tells you what the ride accomplished, what to pay attention to, and what's next.

Races and group rides are handled differently. When Steady detects a race or group ride (from the activity name), it doesn't compare it against whatever was planned. Instead, it evaluates the ride on its own merits — how you performed, what you executed, and what the recovery cost means for your upcoming sessions. If you raced on a rest day, it flags the recovery impact without lecturing you about skipping rest.

The weekly review answers four questions:

1. 1Did your quality sessions deliver what they were designed for?
2. 2Are there patterns worth naming (consistently cutting rides short, improving on long efforts, etc.)?
3. 3What's coming next week and why?
4. 4One actionable coaching note.

It also sees the previous week's review so it can escalate emerging patterns rather than repeating itself. And it draws on your coaching notebook — months of accumulated observations about how you respond to training, what motivates you, and what your tendencies are.

The weekly review can also adjust next week's plan based on what it sees. If you're accumulating fatigue, it might downgrade intensity or reduce session duration for the coming week. If you skipped key sessions, it won't just note it — it'll restructure the week ahead to get you back on track without overloading.

You'll get an email when your weekly review is ready, with a preview of the key insight. Tap the link and it opens directly in the coaching chat.

At the end of each 4-week block, you get a strategic assessment:

• "Did it work?" — Not whether you did every workout, but whether the metrics moved. Did your long ride power improve week over week? Did your heart rate drift decrease? Did threshold tolerance increase?
• "What did we learn about you?" — Coach-level insight: "Volume loaded well but sweet spot sessions struggled — your aerobic base is responding but threshold tolerance needs more gradual loading."
• "What's next?" — Strategic direction for the next block, tied to what the data showed.

RPE (rate of perceived exertion) captures what power data can't — heat, poor sleep, work stress, motivation, how your legs actually felt.

When your perceived effort doesn't match your power output, that's information. High RPE with low power might mean accumulated fatigue or life stress. Low RPE with high power could signal a fitness breakthrough.

After each ride, the coach asks you to rate how the ride felt. Once you submit your RPE, the coach follows up: "Anything else about this ride?" — giving you a chance to add context (flat tire, felt sick, did a group ride instead of the planned session). Your response shapes the review. If you say "no" or nothing notable, the review generates from data alone. If you share something, it gets woven in — "You mentioned work stress this week. That tracks with the RPE-to-power mismatch we're seeing."

Yes, in several ways:

• Coaching notebook — Steady maintains a persistent coaching wiki for each athlete. After every ride review, weekly review, and block review, observations get distilled into the notebook: how you respond to threshold work, whether you tend to skip weekends, what motivates you, your injury patterns, how your body handles volume increases. This notebook is injected into every future review and coaching interaction — so the coach has months of context, not just the current week.
• Each new training block inherits the previous block's peak metrics. Progression starts where you left off, not from scratch.
• The review system tracks your last 2 weeks of execution. If you consistently cut short on Wednesdays or overperform on weekends, the coach notices.
• As your FTP and aerobic threshold change, zone boundaries auto-adjust and the gap analysis re-runs with current data, shifting priorities.
• Each review sees the previous reviews to avoid repetition and to escalate patterns rather than mentioning them once.

The net effect: a coach who started working with you 3 months ago remembers everything from month 1. It doesn't start from scratch each conversation.

A few principles that shape everything:

• Every session needs a "why." Athletes who understand the purpose execute better. "3 hours Zone 2" means nothing. "Building the aerobic base that lets you hold 200 watts at hour 5 without your heart rate drifting" creates buy-in.
• Protect athletes from themselves. Most self-coached cyclists need to be held back more often than pushed forward.
• Reviews should contextualize, not just report. A number without context is useless. Was your RPE-to-power ratio shifting? Is it accumulated fatigue or life stress? The review tells you what the ride meant, not just what happened.
• Handle the messy stuff well. Illness, travel, missed weeks, motivation dips — that's where real coaching earns its value. Not mechanical rescheduling, but intelligent adaptation.

The test we apply: would a great human cycling coach say this? If it sounds like a system notification, we rewrite it.