How Ascend actually works.
No black boxes. Every number Ascend shows you — your elevation, your muscle heat, your strength tier, your readiness — is built from your own logs with maths you can check. Here is exactly how each one is computed.
Workouts → metres of elevation
Your climb isn't cosmetic — it's a direct readout of work done. Each logged set produces a work score from weight × reps, weighted by the muscles recruited. Cardio contributes duration × intensity. The combined score converts to metres of elevation against the mountain you picked, so a heavy session visibly moves your climber upslope.
The chain
- 1. Log a set → weight × reps × muscle-recruitment weight = work score.
- 2. Cardio → duration × intensity adds to the same daily work score.
- 3. Work score → metres, scaled to your mountain's height.
- 4. Climber advances; summits unlock through the five core peaks up to Everest, then optional expansion climbs.
Pick your mountain on the climb page.
Lifts → 30+ muscle regions
Every exercise in Ascend's library is tagged with its primary and secondary muscles. When you log a set, the volume is split across those regions and accumulated over a rolling 7-day window. That per-muscle load is what colours the anatomy heatmap — front and back, across more than 30 named regions — so chest/back/leg imbalances are obvious at a glance, and the strain board surfaces what you've been skipping.
Why volume, not just "did a session"
Hypertrophy responds to weekly volume per muscle group, not session count (Schoenfeld et al., 2017). Ascend attributes set-level volume to each region so the heatmap reflects real dose — a fresh muscle reads hot, a recovered one cools off.
See it in action on the anatomy heatmap.
Bodyweight-adjusted tiers: Recruit → Climber → Summit → Apex
Raw kilos punish lighter lifters. Ascend instead estimates your one-rep max per lift, divides it by your bodyweight, and decays stale PRs so the ranking stays current. The resulting relative-strength figure maps to a tier per muscle — Recruit, Climber, Summit, Apex — refreshed weekly. A 60 kg lifter and a 100 kg lifter can hold the same tier on a level field.
The 1RM estimate (Epley)
estimated 1RM ≈ weight × (1 + reps ÷ 30)
We use the Epley formula (Epley, 1985), cross-checked against Brzycki (1993) for higher-rep sets, then divide by bodyweight for the tier. Try it yourself with the 1RM calculator or check where you land on the bench-press standards.
Sleep + HRV + RHR + load → a transparent 0–100
Most recovery scores hand you a single opaque percentage. Ascend shows the working. Your readiness blends four drivers — sleep, heart-rate variability, resting heart rate, and yesterday's training load — into a 0–100 score, and you can see exactly how many points each driver contributed. Train hard on green days; the coach auto-proposes a deload on red.
Transparent vs opaque
- Sleep — duration + consistency vs your baseline.
- HRV — trend against your rolling average (Plews et al., 2013).
- Resting HR — elevation above baseline flags incomplete recovery.
- Load — yesterday's volume + RPE pressure on today's capacity.
No wearable? Logged sleep and soreness alone produce a score. A connected wearable adds HRV and RHR to sharpen it. See the full breakdown on the readiness score page.
A coach that learns from accept / reject
You pick one of three coach characters — Rae, June or Atlas — and the coach reads your last 90 days of logs before it says anything. Crucially, it doesn't just talk: it proposes actions you can accept, edit, or reject. Every one of those decisions is signal. Your acceptance rate and edits feed the next call, so suggestions converge on what you actually do.
Frames → two actionable cues
Film a working set and Ascend samples frames across the rep to assess depth, bar path, tempo and symmetry. Instead of a meaningless "good job", it returns two specific cues you can act on next set. High-confidence fixes route into the same accept/dismiss flow as the coach, and your clip is analysed privately — never published.
What it checks
Depth, bar path, tempo and left/right symmetry on the main barbell lifts first, with more added as the model is tuned. Details on the form check page.
Frequently asked
How does Ascend turn a workout into elevation?
Every logged set and cardio entry produces a work score — for lifts, weight × reps scaled by the muscles recruited; for cardio, duration × intensity. That work score is converted to metres of elevation against your chosen mountain, so a hard session visibly moves your climber up the slope toward summits like Aoraki, Denali, Everest and the Burj Khalifa skyline.
How are strength tiers calculated?
Ascend estimates your one-rep max per lift using the Epley formula (1RM ≈ weight × (1 + reps ÷ 30)), divides it by your bodyweight, and decays old PRs over time. The bodyweight-adjusted figure maps to a tier — Recruit, Climber, Summit or Apex — per muscle, so a 60 kg lifter and a 100 kg lifter compete on relative strength rather than raw kilos.
Do I need a wearable for the readiness score?
No. Readiness is computed from logged sleep, soreness flags, recent training load and your streak, producing a transparent 0–100 score where every driver's point contribution is visible. An optional Apple Health, Garmin, Whoop or Oura connection adds HRV and resting-heart-rate signal that sharpens it, but it is never required.
How does the AI coach personalise its suggestions?
The coach proposes an action — a next set, a deload, a macro change — and you accept, edit or reject it. Your acceptance rate and edits feed back into the next call, so the coach steadily learns which suggestions you actually take. You pick the coach character (Rae, June or Atlas) up front.
How accurate is the AI form check?
Form check samples frames from a short clip of a working set and returns two specific, actionable cues — for example 'knees caving on the descent, sit wider' — rather than a vague score. It covers the main barbell lifts first, and high-confidence fixes route into the same accept/dismiss proposal flow as the coach.
References
- Epley, B. (1985). Poundage Chart. Boyd Epley Workout — the basis of the 1RM estimate.
- Brzycki, M. (1993). Strength testing: predicting a one-rep max from reps-to-fatigue. JOPERD 64(1).
- Zourdos, M. et al. (2016). Novel resistance-training RPE scale measuring repetitions in reserve. J Strength Cond Res 30(1).
- Plews, D. et al. (2013). Training adaptation & HRV in elite endurance athletes. Sports Med 43(9).
- Schoenfeld, B. et al. (2017). Dose-response of weekly resistance-training volume on muscle hypertrophy. J Sports Sci 35(11).
Go deeper
Start your climb
Free core tracking. 14-day Ascend Club trial.
Join waitlist