The short version: add reps until every set reaches the top of your range, then add weight. Below is everything underneath that, with live figures you can play with.
1strength model for every lift, fitted to your sets
6rules turn that model into a weight and reps
116logged sessions in our first backtest
0ads, trackers or accounts in the app
01
A strength model for every lift
Each exercise gets its own model, fitted to your history on that lift. Every working set gives an estimated one-rep max from its weight, reps and reps in reserve. Those estimates are corrected for fatigue and averaged into a session strength, with more weight on the most reliable sets.
The trend comes from a regression that leans on recent sessions (a 1.5-week half-life) and counts bad days for less. Your weekly rate of gain is blended with a starting point for your experience, and capped at 2.5 times that starting point, so one great week can't run away with your targets.
Fig. 01Estimated 1RM, bench press
From your sessions+0.0%/wk
Starting point+0.5%/wk
Rate used+0.0%/wk
Next session0 lb
Illustrative sessions. Fainter dots count for less because they're older. The ringed dot was a bad day, so it's down-weighted. Switch the experience level to see the starting point pull the trend. Hover a dot for its numbers.
02
Six rules turn the model into a target
Every exercise has a rep range, like 8–12. Defiant projects how many reps you'd get at your current weight, then one of six rules decides what happens next. Press play to watch them fire.
Fig. 02Reps per set, session by session
Your projected reps fit the range
Same weight. Each set matches last time, with at least one more rep overall.
Every set reached the top of the range
Heavier, by the smallest jump your equipment allows that keeps you in the range.
Projected reps fall below the range, and your first set fell short last time too
Lighter, to a weight that fits the range.
A late set would fall below the range, and it fell short last time too
That set backs off the weight.
Three sessions at the same weight with no gain
About 10% lighter, to build back up.
Deload week
About half the sets, about 10% lighter, bottom of the range.
"Your current weight" is the one you used for most sets last time, so a lighter ramp-up set never becomes the baseline. Deload sessions are ignored when working out targets.
03
It adjusts while you train
After each set, the next target is recalculated from the set you just did and how much you personally fade from set to set. That fade starts around 2.5% per set and becomes yours as you log: the median drop between sets at the same weight. Only the set you're about to do shows a target, so it reflects how today is going.
Fig. 03Bench press, 185 lb, range 6–10
How many reps did you get on set 1?
Assumes one rep left in the tank on each set. The app uses the reps in reserve you log, or your usual effort if you don't.
A grinding first set means a lighter second set, so you finish every set inside your range.
04
How well it predicts
We replayed 116 logged sessions and predicted each one using only the sessions before it. Then we compared the model with the simplest rule a lifter might use.
Fig. 04Average reps off, lower is better
Defiant's modelSimple rule
12% closer than "same as last time" on the first set
View as a table
Prediction
Defiant's model
Simple rule
First-set reps, before the workout
2.70 (bias −0.05)
3.08 ("same as last time")
Next-set reps, during the workout
1.06
1.03 ("same as the previous set")
Before a workout, the model beats "same as last time". Mid-workout it matches the best simple rule, and it keeps working when the weight changes. 116 sessions is a small sample. We'll publish bigger backtests as more people train with Defiant.
05
Every target is a weight you can load
You tell Defiant your bar, plates, dumbbells and machine stacks, separately for each gym. Every target, warm-up and weight picker rounds to a load you can make, and barbell lifts show the plates for each side.
Fig. 05What goes on the bar
Machines differ between gyms, so machine and cable lifts keep their history per gym. The first time you use a machine at a new gym, it starts about 10% lighter. Barbells, dumbbells and bodyweight count everywhere.
06
Warm-ups and rest, worked out
Warm-up sets ramp toward your first working weight. Heavy barbell lifts get about 40%, 60% and 80% for 8, 5 and 3 reps, plus a single at about 90% before sets of 5 or fewer. Other compounds get two ramp sets, isolation lifts get one if those muscles are still cold, and muscles warmed by an earlier exercise get one feeder set or none.
Fig. 06Warm-up for a heavy barbell lift
Warm-up setWorking set
Uses the target from Fig. 05, rounded to plates you have. Move that slider and this ramp follows.
Fig. 07Rest between sets
Plus 30 seconds after heavy sets of 5 or fewer, or sets taken to failure. The timer runs on the Lock Screen, in the Dynamic Island and on Apple Watch.
07
Time off is planned for
At most about a week of progress is credited between sessions, however long the gap. After four weeks away, strength is assumed to drop about 1.25% a week, down to a floor of 12% below where you were. Your first session back sets things straight.
Fig. 08Strength assumed after time off
View as a table
Weeks off
Strength assumed
Hover the line to read any week.
08
How programs are built
The program builder asks eight questions: goal, style, days a week, session length, experience, split, gym and focus. Then:
Split. Yours, or the best fit: full body for 2–3 days, upper/lower for 4, upper/lower plus push/pull/legs for 5, push/pull/legs for 6.
Exercises. Each day is a list of movement slots (squat, hinge, horizontal push, vertical pull…). Each slot picks an exercise your gym can do, at your level, that avoids any joint you asked to protect.
Sets and reps follow your goal and the exercise's role. Main lifts are 3 × 6–10 for muscle and 4 × 3–5 for strength.
Time. The most important work goes in first, then more while the day fits your session length. Short sessions pair accessories into supersets.
Cycle. 6 weeks for beginners, 5 for intermediates and 4 for advanced lifters, ending with a deload week.
Fig. 09Weekly hard sets per muscle, before extras stop
CapExtra for focus muscles
View as a table
Experience
Weekly cap
Focus muscles
Beginner
12
18
Intermediate
16
22
Advanced
20
26
Past the cap, extra sets cost more recovery than they return (Pelland 2024). Once every muscle hits its cap, a long session can finish early.
Research
What we build on
Latella C, et al. (2020). Long-term strength adaptation: a 15-year analysis of powerlifting athletes. Journal of Strength and Conditioning Research. Realistic weekly rates of gain by experience.
Pelland JC, et al. (2024). The resistance training dose-response: meta-regressions exploring the effects of weekly volume and frequency on muscle hypertrophy and strength gain. SportRxiv preprint. Weekly volume caps and training frequency.
Bosquet L, et al. (2013). Effect of training cessation on muscular performance: a meta-analysis. Scandinavian Journal of Medicine & Science in Sports. Strength loss after time off.
Helms ER, et al. (2016, 2017) and Zourdos MC, et al. (2016), on reps in reserve. Estimating strength from sets not taken to failure.
Schoenfeld BJ, et al. (2016) and Grgic J, et al. (2018), on training frequency and rest intervals.