iPhone + Apple Watch

FitProofs

Designing for the seconds between sets.

I built the dial that opens the day, the timer you check between sets, and the visual language that makes them feel like one product.

FitProofs Home: the daily dial and a saved workout in pearl.
FitProofs rest view: the countdown, next set and logged sets in charcoal.
Role
Product designer and builder
Scope
Workout flow, Body profile, food logging
Focus
Glanceability, editable estimates, accessible glass
Scroll to explore

From a single glyph to a workout on the wrist.

I wanted the app to feel like FitProofs even when only one number fits on the screen. The daily dial, the rest timer and the Watch set editor share the same type, surfaces and motion. The atomic library documents 60 native components against 24 iPhone and 19 Watch captures.

The brief

Don't make someone operate two screens during a set.

The phone is a good place to choose a workout. Mid-session, it is often on the floor. I wanted the wrist to answer the immediate question without asking someone to pick the phone up.

My contribution

Decide what belongs at a glance.

I set the hierarchy, moved body exploration out of Settings, and directed the light and charcoal treatments. I also kept a manual review step where a sensor or photo could give the wrong answer.

The part of the workout that happens on the phone.

The wrist is good at a glance. The phone is where I can check a portion, correct a set, or look back through a day. These four journeys use native simulator captures with example data, not a personal activity log.

Ready screen for a dumbbell bench press, with 10 target reps, a 24 kg working load and a set marked Not logged.
iPhone · Training / Prepare

The task

What do I need for this set?

Design decision / 01

A target isn't a result.

The ready screen puts the target reps beside the working load. The set below still says “Not logged”. I wanted a plan to be useful without letting it quietly become a workout record.

View full-size capture

What I missed until I wore the Watch.

These are findings from using my own builds, not a study of other lifters. For each one, I wrote down what happened before deciding how the design should respond.

01 / Cross-device continuity

Two readable clocks can still give the wrong answer.

I added rest on my iPhone and the Watch kept counting down to the old time. Looking at either screen alone, nothing looked broken.

Design decision
Treat rest as one deadline shared by both devices, not as two independent countdowns.
How I'd retest it
Extend rest on the phone, let the Watch reconnect and compare the end times. Neither device should restart the set.
02 / Health data

Missing is not the same as zero.

Steps and sleep updated while active calories did not. A general “Health connected” status couldn't tell me whether this particular reading had arrived.

Design decision
Distinguish an unavailable reading from a measured zero, and keep workout energy separate from the day's total.
How I'd retest it
Follow one workout from Apple Health to the live session and then to Home. Check the source value, time window and unit at each stop.
03 / Navigation

A shoulder shouldn't be a Settings task.

I kept going through Settings to reach Body. Once there, a selector asked me to name a muscle I could already point to.

Design decision
Give Body a direct entry, keep the app tabs in place, and make the illustrated shoulder the selection target.
How I'd retest it
Start on Home, inspect the shoulder and return to training. Watch for the first hesitation; don't prompt the person where to tap.

Working on a similar handoff between devices? Discuss a mobile UX project.

Take the glass away. Can I still read my meal?

Here is the same meal editor in charcoal, with the same example portion and macros. In the solid treatment the boundary becomes more distinct. The food name, gram control and four nutrient values keep their positions, so reading the solid version doesn't mean learning a second layout.

Standard glass: close-up of the salmon portion control and four nutrient values in the dark meal editor.
Standard glass

Tint and a soft edge separate the food from the meal sheet.

Reduced transparency: close-up of the salmon portion control and four nutrient values in the dark meal editor.
Reduced transparency

The solid surface and outline separate the same controls without relying on translucency.

Accessibility in the design library

The shoulder is the selector.

I removed the playback controls and asked for a direct tap on the body. Blue fibres identify the selection; a second, smaller view makes its shape easier to inspect. These are two captures from the charcoal design iteration.

Front body illustration with muscle selection targets.
01 / Orient
Right shoulder in blue, enlarged beside a detail inset.
02 / Inspect

Three moments in one set.

The number that matters changes during a workout. In a working set it is the rep count. Between sets it is the remaining rest time. In review it is the count you can correct. These are simulator captures with example data.

FitProofs on Apple Watch: Lift
01 / Lift

Keep the exercise and set in view.

FitProofs on Apple Watch: Rest
02 / Rest

Give the remaining time the strongest contrast.

FitProofs on Apple Watch: Review
03 / Review

Let the lifter correct the set before saving.

Browse the screen atlas

An estimate the lifter can correct.

What shipped

  • iPhone workout plans and food logging with a review step.
  • A Watch workout view that distinguishes the rep target from the set you log.

What was measured

Offline rep-count comparison
On 439 complete development sets with known boundaries, 46.2% of the current counter's estimates and 82.0% of the research model's were within one rep.

These are not live Apple Watch accuracy figures. The research model was not added to the Watch app on the strength of this benchmark.

The call I’d revisit

I'd test the recovery moment first.

A lifter holding a partial rep and one resting between sets can both look still. Before adding another automatic transition, I'd test what happens when the Watch guesses wrong and the person needs to recover.

What I’d instrument next

  • Compare confirmed counts against consented Watch recordings, including actual rests and transitions.
  • Check whether someone can find Body, inspect a shoulder and return to their workout unaided.

Notes from building FitProofs.