Layers

So the app keeps the record for you, and asks for almost nothing in return. You photograph the outfit (a mirror shot, two seconds) and Layers freezes the day’s weather onto it: the temperature, the high and low, the sky, the place. That reading never changes. Months later the entry still remembers the day it belonged to, not today’s.

Some mornings there’s no photo, so there’s a second way in: a set of hand-drawn garment icons (t-shirt, long sleeve, shell, boots) that assemble an outfit as fast as you can tap. Either way the whole interaction weighs one photo or a few taps, because a diary you have to think about is a diary you stop keeping.

In the evening it asks a single question, on a scale from too cold to too warm: how did it feel? That one tap is the hinge of the entire product. It’s what turns a photograph into something the app can reason about later, and it’s the only real work the app ever asks of you.

I wanted the opposite of a weather app. The category is loud: saturated gradients, urgent alerts, numbers stacked on numbers. None of it is calming, which is the one thing you want from the screen you check before you’ve had coffee. Layers is built on warm paper instead of black glass, sets its headings in a quiet serif against a plain humanist sans, and lets each screen hold a single idea.

Weather still has to be felt, so it’s carried by colour rather than ornament. Every condition owns a soft wash (sun, sky, rain, sage, lavender), and those washes run through the whole app: behind the temperature, across each saved look, as the tint of a day in the calendar. The sky itself is drawn, not stock: a canvas that blooms, clouds over, rains and turns to a moon-and-stars night, tuned to read as atmosphere and never as a screensaver.

The rest is the small, tactile work that decides whether a thing feels made. An icon-in-a-circle motif repeats everywhere. Controls give a little under the thumb, and the day’s temperature is one calm spline instead of a jagged chart. It’s the same instinct as my hardware work, that an interface should feel like an object with a right way to be held, pointed at a screen instead.

The point of all that logging is one moment: you open the app and it shows you what worked. Layers looks back through your rated outfits, finds the past day whose weather most resembles today’s (temperature counts for the most, then the sky, then the day’s spread) and returns the one you rated best, but only if it’s genuinely close. The reason is always in plain words: you wore this on a similar 18–26° day and rated it perfect.

What I care about most is what it does before it can do that. A recommendation with three data points behind it is a liar, so under its threshold Layers doesn’t pretend to personalise. It says what it can honestly draw from the forecast alone (cool morning, warmer afternoon, layers may help) and tells you it’s still learning your style rather than dressing a guess up as a result. The logic stays deliberately simple and always explainable; I’d sooner it be right and modest than clever and wrong.

That honesty carries a cost, and it’s the hardest design problem in the app: the payoff only arrives once you’ve logged enough, so the first week is the week Layers is least useful and most easily abandoned. Most of the recent work has gone into that gap: a visible “calibrating” goal that gives the first days an end, and a way to seed history from photos you’ve already taken, so the app earns its keep sooner without ever faking the part that has to be earned.

And it doesn’t wait for you to open the app. The same recall rides the Home and Lock Screen as a widget, a morning glance that says you wore this on a similar day and it worked, so the one useful moment lands where you already are at eight in the morning.