FoldcardFor iPhone Duo · free, with a $4.99 unlock
Stand it up.Revise with both hands free.
Bend the device to about 110°, set it on the desk, and the session splits in two. The card sits on the upper panel at reading angle. The four grade buttons sit on the lower panel under your fingers. One fold per session, not four hundred.
That is the real scheduler. Press Good four times and you get 10m, 1d, 2.5d, 6.25d — the same ladder the app's tests assert. Tap the card to reveal, or use space and 1–4 on a keyboard, exactly as in the app.
What the hinge is for
The fold sets the posture once, and is then left alone.
A flashcard app whose headline feature is folding the phone to flip each card is the one idea research rejects outright. A person doing spaced repetition performs 200 to 400 reviews in a sitting; a fold is two hands, a change of grip and a deliberate motion. Foldcard never flips a card with the hinge.
One fold per session
Stand the phone up and the layout splits. That is the whole of it. Nothing inside a session depends on the hinge at all.
Reading angle and typing angle
The card is where you look, the grades are where your fingers are. It is the posture of a laptop rather than of a phone, and nothing else in a pocket does it.
A fold never grades a card
Not at any angle, not at any speed. The posture classifier's entire output is a layout; there is no code path from it to the scheduler.
Why dragging the hinge above feels deliberate
A hinge stays exactly where you leave it, and a raw hinge angle is noisy and quantised. Three things stand between the sensor and the screen, and you can feel all of them in the control above:
| Anchor and band | 110° with ±20° around it. The full 0–180° sweep is deliberately not mapped — nobody studies at 179°. |
|---|---|
| Hysteresis | Leaving a posture takes 8° more travel than entering it, so a hand resting on the boundary does not flip the layout back and forth. |
| Debounce | A posture must hold for 400 ms to be adopted. Flick the hinge quickly across the band above and nothing happens — that is picking the phone up and putting it down again. |
On a phone without a hinge you get the single-panel layout and the complete app. The posture is a comfort, not a dependency.
Why the app exists
Bring the deck you already study.
Import an Anki package and keep your scheduling history, your intervals and your media. A new study app with no content has no reason to exist, so import is the first-run path rather than a settings item.
Real numbers, from importing a package Anki itself wrote. Not illustrative figures.
Nothing is dropped quietly
The report names every conversion and every card that did not come across, with counts. A student importing four years of reviews needs to see that number before they trust the app with a session.
Suspended stays suspended
A large shared deck is mostly suspended cards. They are imported, counted, and kept out of your sessions — rather than handing you thirty thousand cards on day one.
Sub-decks keep their names
Course::Chapter 2 arrives as its own deck with its full name,
so two chapters of the same course stay apart.
The scheduler
SM-2, implemented plainly and written down.
Not a novel algorithm. SM-2 is the one behind classic Anki: predictable to anyone arriving from it, and — the part that matters — it keeps an imported deck's existing intervals meaningful.
Log scale. Computed live by the same code as the demo above.
| Starting ease | 2.5 |
|---|---|
| Minimum ease | 1.3 |
| Again / Hard / Easy | −0.20 / −0.15 / +0.15 |
| Learning steps | 1 minute, 10 minutes |
| Graduating interval | 1 day |
| Easy interval | 4 days |
| Daily limits | 20 new, 200 reviews — adjustable |
| Day rolls over at | 04:00, not midnight |
Every value is listed in the app's own Settings, read-only. Intervals carry a deterministic ±5% spread so a ten-thousand-card import does not pile every card onto the same future day — the same import twice gives the same dates.
Built for long sessions
The question stays on screen while you grade the answer.
The answer appears below a rule
It does not replace the question. Losing the question at the moment of judging the answer is the most common complaint about phone flashcard apps.
Taps only, and the buttons never move
No swipe gestures — they are error-prone at speed and inaccessible. Good is the majority answer, so it is the emphasised one, and each button prints the interval it will give you before you press it.
Right-to-left, per field
A Hebrew–English deck has an RTL front and an LTR back on the same card. Direction is detected per field, not per card. Two cards in the demo above show it.
Text up to 34 point
The single most common request in every flashcard app is bigger text, and Dynamic Type scales on top of that, all the way to the largest accessibility sizes. The grade row reflows to 2×2 rather than clipping its labels.
Keyboard and VoiceOver
Space reveals, 1–4 grade, Escape leaves. VoiceOver reads the card and announces each grade with the interval it will schedule.
No streaks, no XP, no leagues
Spaced repetition already has a feedback loop, and bolting a second one on damages it.
Plainly
What it does not do.
There is no sync.
Your cards live on this device only. If you study on a laptop and a phone and need them to match, this is not the right app — and we would rather say so here than in a refund.
There is also no account, no sign-in, no analytics and no networking code of any kind. Nothing is collected because nothing is sent anywhere. The App Store privacy label reads Data Not Collected, and it is accurate rather than optimistic.
No subscription, at any tier. Free for three decks with unlimited cards in each; one payment of $4.99 removes the deck limit and adds the card editor. The cap is on decks and never on cards, because the common case is one large imported deck.
A study station that fits in a pocket.
Foldcard arrives with iPhone Duo.
App Store link coming at launch
Works on any iPhone running iOS 26 or later. Without a hinge you get the same app in a single panel.