On track
Did every session through week 6 on target. It is now Monday of week 7 (week 7 has not started).
On track - no changes needed.
Auto Train rebuilds your plan around the life you actually ride: every missed session, every breakthrough, every week off the bike. It is grounded in published sports science, every change is explained in plain language, and its behaviour is verified across more than two thousand automated tests, so you can always see exactly how and why it adapts.
Three ideas make the whole thing feel like a coach, not a spreadsheet.
Fitness is the training you have banked over the last six weeks. Fatigue is how tired you are right now. Freshness is the balance between them, how race-ready you feel. Every ride you do moves all three.
Auto Train never makes you "catch up" on missed work. It reads your true current fitness and rebuilds the road to your goal from there, at a pace your body can absorb. No backlog, no doom spiral.
Every change is explained in plain language and waits for your tap. Nothing happens to your plan without you, and anything can be undone. You are always the one driving.
None of this is invented. Auto Train encodes the same principles coaches and exercise physiologists have refined over decades, and it stays deliberately conservative where the science is uncertain.
Fitness only grows if you add load, but add it too fast and you break down. Auto Train grows your training at a safe ceiling that scales with your current fitness, and eases automatically for riders who are new, returning, or older.
Progressive overload · athlete-specific ramp capEvery third or fourth week is a planned lighter week so your body can absorb the work and come back stronger. New and masters riders recover more often. It is structural, never skipped.
Deload periodization · 3:1 and 2:1 cadenceIn the final two weeks Auto Train cuts volume while holding intensity, the taper shape shown to deliver the biggest race-day gain, so you reach your event sharp instead of stale.
Bosquet taper meta-analysisOne low form reading is not enough to force a change. Auto Train only eases hard when signals agree, and it refuses to let a re-plan dig your fatigue below a safe floor.
Meeusen consensus · convergence + fatigue floorA wrong FTP corrupts every zone and target. Auto Train raises it only from a real, measured effort, and it will never quietly lower it. If your numbers slip it invites you to re-test, on your terms.
FTP integrity · eFTP detectionPeaking for an event, building general fitness, coming back from time off, or just maintaining, your goal scales the whole plan and every future adjustment, so it always fits why you ride.
Volume · intensity · ramp scaling by goalEach card below is computed live, right now, by the exact engine that runs your plan. These are the real, deterministic decisions Auto Train makes, for the moments that actually happen in a season.
You did the work. The best adaptive system knows when to leave a good plan alone.
Did every session through week 6 on target. It is now Monday of week 7 (week 7 has not started).
On track - no changes needed.
Rode weeks 1-7 exactly as prescribed. Evaluated on the Sunday of week 7, when fatigue (TSB) bottoms out.
On track - no changes needed.
On plan through week 6, but ONE session came in easy (60% of target). A single off day is not a signal.
On track - no changes needed.
On plan every week, plus the same extra weekend ride week after week. Auto Train treats that habitual ride as part of your real training (it already counts toward your fitness), so it stops nagging you to adjust for it. A one-off big week would still get flagged; a reliable habit will not.
On track - no changes needed.
One missed ride is noise, not a signal. Auto Train protects the week without over-reacting.
On plan through week 6 except ONE skipped session in week 6. Now Monday of week 7. One miss is noise, not a signal - the plan holds.
On track - no changes needed.
On plan through week 6. This week you did everything except Monday's EASY session; it's Thursday. A missed easy ride is a non-event - rest already does its job - so Auto Train leaves the plan alone. It should NOT be shuffled onto a rest day.
On track - no changes needed.
This week has two quality sessions (Tue/Thu); you missed Tuesday's KEY session and it's Thursday. Losing it leaves only one quality session, so Auto Train moves it to a healthy open day - keeping a recovery day either side (never onto the day beside another hard day).
Moved your missed Threshold - Key Session (a key session) to Sat, keeping a recovery day either side. The rest of your plan is unchanged.
Moved Threshold - Key Session: Tue → Sat
Missed Tuesday's KEY session and it's now Sunday with no open day left. Auto Train absorbs it rather than cramming a hard session in under-recovered - one clean session beats two compromised ones.
On track - no changes needed.
On plan through week 6, but it is now WEDNESDAY of week 7 and the Mon/Tue sessions were skipped.
Adjusted upcoming sessions (2 missed).
On plan through week 6 except TWO skipped sessions in week 6. Now Monday of week 7. Two misses is a signal → a small re-sequence.
Adjusted upcoming sessions (2 missed).
Come back from a real break and the plan rebuilds around the fitness you actually have now, not the one you left.
On plan until a bug cost you the first sessions of this week; it's now Thursday. Auto Train reads this as lost days, not a drop in your ability, and simply eases you back on. Rest is the priority.
Adjusted upcoming sessions (2 missed).
Did weeks 1-5, then skipped ALL of week 6. It is now Monday of week 7.
Several missed sessions - plan rebuilt from your current fitness.
Did weeks 1-4, then skipped weeks 5 AND 6 entirely. Now Monday of week 7.
Several missed sessions - plan rebuilt from your current fitness.
Only weeks 1-3 done, then a long gap - weeks 4, 5, 6 all skipped. Now Monday of week 7.
Several missed sessions - plan rebuilt from your current fitness.
Missed 3 scattered sessions back in weeks 2-3, then rode weeks 4-11 exactly. It is now week 12. Those old misses are already reflected in your (decayed) fitness, so they must NOT rebuild the plan - the reactive signal is windowed to recent training.
On track - no changes needed.
Did every session for six weeks, but rides come from a heart-rate strap with no power meter. Auto Train estimates the training load from your heart rate, so your sessions count and your fitness is real. It never reads a compliant rider as if they missed everything.
On track - no changes needed.
Rides every session to prescription, HR-only - and the HR estimate reads a median 0.81 of what power would say, so every quality session lands around 81% of target. The power bar (0.8) called that a coin flip per session, and two flips meant 'eased the block - re-check your FTP'. Provenance-aware grading judges an HR estimate against the HR bar: every session completes, nothing is blamed on the athlete.
On track - no changes needed.
Not one bad week, but a pattern: falling short of the plan, or reliably riding it at the wrong intensity. Auto Train notices the drift and helps you re-align.
Weeks 1-2 done fully, then weeks 3-6 you kept showing up but kept coming up a session short, powering through the smaller nudges along the way. That steady, spread-out shortfall (not one bad week, not a total gap) is exactly what Auto Train stops nagging you about week by week. Instead it makes one clean call: rebuild around the training you are actually doing.
You have completed about 78% of your planned sessions over the last 4 weeks. Rather than nudge single weeks, Auto Train rebuilt your plan around the training you are actually doing, still aimed at your goal.
Completing ~78% of planned sessions
Rode weeks 1-6, but session after session kept landing around 70% of target, under the 80% "completed" line, and you kept pushing through. One under-target ride is nothing; a steady run of them is a real signal, so Auto Train eases the block back to a load you can actually hold.
Repeated under-target sessions while fresh - eased the block, and it's worth re-checking your FTP.
Suggests a fresh FTP test (never lowers it for you)
Rode weeks 1-6 and hit every session's load, but did them all as hard, punchy efforts even on the easy aerobic days. Auto Train credits the sessions (you trained, and the load is right, so no phantom misses) but notices you keep riding harder than prescribed and flags the drift, so your easy days can go back to being actually easy.
On track - no changes needed.
Rode harder than planned on 8 recent sessions
Get stronger and the plan follows. Overdo it and it reins you in. And it will never quietly lower your FTP, only invite you to re-test.
On plan through week 6, PLUS a recent ride that set a 20-min best of 305 W (current FTP 250).
Adjusted upcoming sessions (extra riding logged).
FTP looks higher: 290W (from 250W)
A steady 20-min effort a few weeks ago set a 285W baseline, but a recent steady 20-min came in at 253W - a genuine drop below the athlete's OWN recent ceiling (not just below the set FTP). Auto Train suggests a fresh test; it never lowers your FTP for you.
Adjusted upcoming sessions (extra riding logged).
Suggests a fresh FTP test (never lowers it for you)
On plan through week 4 (a recovery week), PLUS an unplanned 90 TSS ride that Saturday. It is now Monday of week 5.
Adjusted upcoming sessions (extra riding logged).
You felt strong and stacked two extra endurance rides on top of a full week. Auto Train debits a little upcoming load so a great week does not snowball into too much fatigue, too soon.
Adjusted upcoming sessions (extra riding logged).
Skipped weeks 5-6 (→ plan rebuild) AND set a 20-min PR of 305 W (→ FTP suggestion). Both must surface.
Several missed sessions - plan rebuilt from your current fitness. (extra riding logged)
FTP looks higher: 290W (from 250W)
Ride with Garmin, Wahoo, or Strava and every session still counts. Imported rides credit their planned slots, off-plan imports count as extra load, and a ride recorded twice is never counted twice.
Did every session through week 6 on target, but the rides all synced in from an external platform (Garmin/Wahoo/Strava FIT files) rather than being ridden natively. Auto Train must match each imported ride to its planned slot and credit it exactly like a native ride - a compliant rider is never read as missing everything just because the data came from a FIT import.
On track - no changes needed.
On plan through week 4 (a recovery week), PLUS an unplanned 90 TSS ride that Saturday that synced in from an external platform. An imported off-plan ride must be read as extra load (it counts toward fitness), not as a missed session.
Adjusted upcoming sessions (extra riding logged).
On plan through week 6, but one session was recorded natively AND also synced back in as a FIT import - the same physical ride, two records (same start minute + duration). Auto Train must collapse the duplicate: it counts the load once and never surfaces the second copy as a phantom extra ride.
On track - no changes needed.
We check Auto Train at two levels. First, the mechanics: 2,112 generated plans and the 28 real situations above run through the exact engine on every change, and it is deterministic. The same inputs always produce the same decision, so its behaviour is reproducible and auditable, not a black box.
Second, and this is the rare part: we simulate whole training seasons. A virtual rider rides the exact engine you will, week by week, through everything a real season throws at you: a missed week, an illness, a breakthrough, the standing weekend group ride. Then an independent fitness model the engine never sees checks that each plan genuinely built fitness and arrived fresh for the event, safely. Over a hundred simulated seasons, every one reproducible. It is validation in simulation, not a human trial, but it is a long way past hoping the plan looks right.