How does Kiwi pick my weights?
Kiwi starts from what people like you lift: your sex, size, age and experience. The first time you do an exercise, your sets climb from light toward your likely weight. After that, every workout you log moves the next one: a step up when you get every rep, the same or lighter when you miss reps, and your own weight when you pick a different one. You never test a one-rep max or rate how a set felt.
Where do my first weights come from?
From Kiwi's own data. Before you log anything, Kiwi knows only what you told it when you signed up: your sex, height and weight, age, how much you've trained, and any injuries. It compares that with what Kiwi members like you lifted on their first workouts, which show how strong people of each sex, size and experience really are when they start, and how exercises compare with each other, like a dumbbell press with a bench press or a leg press with a squat. We use members' workouts only as totals and averages.
Two public sources fill in what our members' logs can't show on their own:
- OpenPowerlifting. Its public-domain records of about 950,000 drug-tested competition results show how strength grows with body weight and fades with age, and how women's and men's strength differ.
- Published studies. Studies that measured the same people on two exercises cover the exercises few members have done yet.
Size counts through muscle, not total weight. If you've done a body scan, Kiwi uses your lean mass; if not, it estimates lean mass from your height and weight. Every weight is rounded to plates, dumbbells and machine stacks gyms actually have, in pounds or kilograms.
One thing our members' data showed us: on their first workouts, people squat about what they bench and deadlift only a little more, while competitive powerlifters squat about one and a half times their bench. So Kiwi starts your leg lifts where people actually start, and moves them up as you show it you can do more.
Why does my first workout start light?
Because nobody knows your strength yet. Two people with the same sex, size, age and experience can be far apart: among people who match your profile, the strongest quarter lift at least 70% more than the weakest quarter. So any single guess would be too heavy for a lot of people. Instead, the first time you do an exercise, your sets climb. The first set is light for nearly everyone like you, and each set after it steps up toward your likely weight, so the workout is a warm-up and a test in one. If a set gets hard, stop climbing and stay there.
Kiwi also leaves more in the tank if you're 60 or older, and on exercises that load an injury you told it about.
- 45–50 lbFirst time: the sets climb
- 60 lbEvery rep: two steps up
- 70 lbEvery rep: two steps up
- 80 lbMissed reps
- 80 lbSame weight, every rep
- 85 lbEvery rep: a step up
- 70 lbBack after 3 weeks off: lighter
How do my weights change after a workout?
- You got every rep: next time is a step heavier: the next plate, dumbbell or pin. If your first weights were cautious and you keep finishing them, Kiwi moves up two steps at a time until it catches up with you.
- The next dumbbell is a big jump: going from 10 to 12.5 lb is a quarter heavier, so Kiwi adds reps first and moves you up once you reach the top of the range.
- You missed reps: the weight holds, or comes down to what you did in full. Miss the same weight twice and it drops about 10%.
- You picked a different weight: Kiwi goes from yours.
- You took three weeks or more off: you start lighter. Kiwi assumes strength fades about 1% a week after three weeks off, up to 10%.
A logged weight that's wildly off from your others, like an extra zero or kilograms entered as pounds, doesn't change anything until your next workout says the same.
Does one exercise tell Kiwi about another?
Yes. Kiwi keeps one picture of your strength: overall, in your upper and lower body, in each muscle group and on each exercise. Log a bench press, and your dumbbell press and incline press move a lot, your rows a little and your squats barely. So when the coach adds an exercise you've never done, it starts from what you've already shown, and the sets still climb the first time.
Do I have to rate how hard each set was?
No. Kiwi reads what you log: the weight and the reps. If a set hurt or felt wrong, tell the coach in chat and it will change the exercise or the weight, since pain is something your logs can't show.
How well does it work?
We tested it on 2,098 exercises that members logged in Kiwi (more than 6,500 sets), looking only at totals. Each suggested weight was checked against what that person actually lifted for those reps in the next three weeks, and counted as too heavy if it was more than one step above all of it (5 lb on a barbell). Each member's weights were predicted from standards built without their own workouts.
| When | Suggested weights that were too heavy |
|---|---|
| The first workout | 5.4% (the first set: 1.3%) |
| An exercise that's new to them | 5.1% |
| The next workout | 5.7% (repeating their last weights: 7.9%) |
| Back after three weeks or more | 8.4% (going straight back to their best recent weights: 34%) |
We also checked our standards against research we didn't use to build them. In our data, members who already train start 1.6 times as strong as beginners, and those who call themselves athletes about twice as strong; four studies that tested trained and untrained men on the bench press found 1.7 to 2.2 times.
The cost is caution: on the first workout, most suggested weights are lighter than what people pick for themselves. Finish your reps and Kiwi moves you up quickly.
Want a plan that does this for you? Get Kiwi, free to download on iPhone and Android.
Sources
- OpenPowerlifting. Competition results, data release of October 17, 2026. Public domain. openpowerlifting.org
- Boer P. Estimated lean body mass as an index for normalization of body fluid volumes in humans. Am J Physiol. 1984;247(4 Pt 2):F632–F636. PubMed (lean mass from height and weight)
- Jaric S. Muscle strength testing: use of normalisation for body size. Sports Med. 2002;32(10):615–631. (how strength grows with body size)
- Brzycki M. Strength testing: predicting a one-rep max from reps-to-fatigue. JOPERD. 1993;64(1):88–90; and Epley B. Boyd Epley Workout. 1985. (reps and the one-rep max)
- Same-person comparisons between exercises from 84 published studies and theses, among them: Rodríguez-Ridao D et al. Effect of five bench inclinations on the electromyographic activity of the pectoralis major, anterior deltoid, and triceps brachii during the bench press exercise. Int J Environ Res Public Health. 2020;17(19):7339 (doi); Ferrari R et al. Muscle mass and training status do not affect the maximum number of repetitions in different upper-body resistance exercises. Open Sports Sci J. 2017;10:81–86 (doi); Tsoukos A, Bogdanis GC. Effects of transducer placement on load–velocity relationships in Smith machine and free weight squats in trained women. J Funct Morphol Kinesiol. 2025;10(2):178 (doi); Nieto-Acevedo R et al. Should we use the men load-velocity profile for women in deadlift and hip thrust? Int J Environ Res Public Health. 2023;20(6):4888 (doi); Androulakis-Korakakis P et al. The minimum effective training dose required for 1RM strength in powerlifters. Front Sports Act Living. 2021;3:713655 (doi)
- Nuzzo JL, Pinto MD, Nosaka K, Steele J. Maximal number of repetitions at percentages of the one repetition maximum. Sports Med. 2024;54(2):303–321. doi:10.1007/s40279-023-01937-7 (its dataset: the trained and untrained bench press comparison)