A design study · Foldables
Most banking apps pretend the phone has one screen.
Samsung has shipped foldables for seven years. Apple's first is expected this year. Almost every banking app answers the open canvas by stretching one column across it. I built Meridian, a hypothetical bank, to work out what should actually happen when the phone opens. Four devices, three postures, one live prototype you can fold.
Unfolding changes what the screen is for, not just how big it is.
If a layout would work identically as a stretched phone screen, it was the wrong layout.
That rule is easy to assert and harder to defend, so here is the thing itself. Press Open and watch what the screen becomes.
Loading the prototype…
The cover screen masks money by default.
The cover screen is the only display on a phone that faces outward while the device sits on a table, a desk, a café counter. It is the most publicly visible screen most people own, and it is the one screen the industry treats as a glanceable dashboard for your balance.
So in Meridian the cover shows the shape of your finances without the numbers. Account names, structure, and a masked total. What left the account today, masked. Enough to know something happened. Not enough for the person opposite you to read. Revealing is one tap, and Amounts sits as a permanent item in the navigation rail rather than buried three levels into settings, because a privacy control you cannot find is not a privacy control.
The fold introduces a new surface, and new surfaces inherit old assumptions unless someone stops to check them. A balance that was private on a phone in your hand is not private on a cover screen on a table.
Before layout, hardware.
Most foldable concepts assume the open canvas is a tablet in portrait. It is not.
| Device | Cover canvas | Open canvas | Hinge | Crease |
|---|---|---|---|---|
| Book-style, current gen | 400 × 933 | 856 × 950 | Vertical, book | 24 px continuous |
| Book-style, expected | 440 × 956 | 888 × 956 | Vertical, book | 8 px continuous |
| Clamshell, square cover | 344 × 380 | 380 × 887 | Horizontal, clamshell | — |
| Clamshell, square cover | 352 × 355 | 380 × 887 | Horizontal, clamshell | — |
The cover width survives as pane one.
The open layout is not a redesign, it is the cover plus a second pane. That is what keeps this implementable rather than theoretical.
The crease is specified per device, not assumed. Nothing crosses it. Not a number, not a control, not a chart line.
The shell, closed and open
Five destinations throughout. Home, Invest, Send, Services, Rewards.
Closed, they are a bottom bar. Open, the same five become a left rail with Search, Amounts, Alerts and Profile joining them at the edges. Identical labels, identical order, identical icons.
This is the rule that makes a foldable feel like one app rather than two. Users who rebuild their mental model every time they open the device stop opening the device.
Opening or closing is a layout change, not a user decision. Selection, input, scroll position and the current step all survive it. Break that once and the fold becomes a cost rather than a feature.
Most flows keep the primary task in pane one and use pane two for supporting context. Home does not.
Home is for scanning and launching, not for selecting an item and viewing its detail. No single block deserves to own the second panel. So it reflows into one coordinated dashboard instead of splitting.
Accounts, quick actions and recommendations hold the left. Activities and transactions fill the right. Same content, same actions, same components. Only hierarchy and placement adapt to the wider canvas.
Restraint took longer to write than the layouts.
The test. Does keeping related content visible materially improve the task? If not, the larger canvas stays focused rather than manufacturing a second pane.
Worth designing for
- Send. The payee list stays visible while the selected transfer opens beside it. List plus action.
- Invest. One phone screen answers "what would this fund do." Two panes answer "what would these three do, with my money, over my timeframe." Comparison is the thing a folded phone physically cannot do.
- Cards. Card, balance and controls hold still while the statement scrolls beside them.
- Home. The exception above.
Leave it alone
- Login and biometrics. Contained and single focus at every width. Expanded width should never expand the security surface.
- Checking a balance. The cover screen already answered it. Requiring an unfold is a regression.
- OTP and PIN. A capped sheet over the current task, never an expanded layout. Small things stay small.
- Confirmations and receipts. One clear surface. Do not add filler to occupy the width.
- Onboarding forms. Linear and capped. Guidance may sit beside the fields without splitting their sequence.
Two days, and what I would test next
The speed did not come from a tool writing code quickly. It came from spending the first hours on decisions and then writing them down properly.
Staged and checked beat one large generation every time. The design judgement was still the bottleneck and still the differentiator. The tool removed the execution tax, not the thinking.
Protective or friction
Whether masked by default reads as protective or as friction. My assumption is protective. It needs a real person and a real café table to confirm.
Reachability, two handed
Reachability on the open canvas held two handed. The right pane sits near the thumb. The left does not.
No invented metrics. This study has no users, so the output is a rule set rather than a percentage.
Those are the rules the layouts came from. Go back to the prototype and see whether they hold.
Sources. Android adaptive layouts and canonical patterns · Android postures and hinge occlusion · Samsung One UI large screen guidance · Samsung app continuity guidance · published device specifications, current generation.
