A notebook that tests you on what you wrote. An iPad notebook for students. Mark what matters while you write, and it comes back as a question before you forget it.
Role: Solo product designer
Scope: Research, UX, UI, a native iPad app and its landing page
Platform: iPad, iPadOS 17 or later
Status: In TestFlight beta, 13 testers
My role. Solo product designer. I used Claude for research, checking and process automation, and to write the code. Every design and technical decision, every test and every change was mine; AI helped me carry them out.
01 · THE IDEA
I was learning the app instead of the subject I came to study.
As a student I tried many digital notebooks, and each one took time to figure out. After a while I'd realise I wasn't studying: I was learning the app's features. So I went back to paper, where everything was in front of me and I could mark what mattered and bookmark what I needed.
Not long ago some friends of mine started university, and the question came up again. What I didn't make for myself back then, I decided to make for the students who come after me.
The principle: Don't get pulled away from studying.
Everything you need is in front of you from the first time you open it. Pages run as one continuous strip, with an empty page always waiting below. There is no "next page" button: a student in a lecture shouldn't be distracted by extra controls.
02 · UNDERSTANDING THE PROBLEM
What works for memory, not what a few people prefer.
I built TakeNotes on published research. Twelve university studies give exact numbers on which study methods work; a handful of interviews would have told me what a few people prefer. I also had my own years as a student, and friends who just started university: many struggle with revision, and not everyone knows how to study well.
Real: 12 published studies; a market scan of Goodnotes, Notability, Quizlet and Anki.
My own: My years as a student; conversations with friends who just started university. Informal, not recorded.
AI-simulated: Three proto-personas built from the studies; a usability test with 8 simulated testers.
Three findings, and what each one changed
Finding 1 · Testing yourself beats rereading.
Finding: A week later, students who tested themselves remembered 61%; students who reread, 40%. Most test themselves late, only to check.
So: The notebook has to ask the questions, and keep asking.
Decision: Whatever you mark becomes a question, and comes back when you're closest to forgetting it.
Finding 2 · Students study around deadlines.
Finding: 66% cram the night before a test; only 13% plan ahead.
So: Use the date they already care about.
Decision: Add an exam date, and the app spreads the questions you're unsure about over the days left.
Finding 3 · A highlight alone doesn't help you remember.
Finding: Highlighting and rereading are rated low-utility; practice testing, high.
So: What you mark should become something you're tested on.
Decision: Test myself covers every marked part of a page; you write the answer on top, then tap to check.
Sources: Roediger & Karpicke 2006 · Hartwig & Dunlosky 2012 · Dunlosky et al. 2013.
Proto-personas. Built from the studies, not from interviews; every trait carries its finding.
User flows. Six key flows, tagged by persona.
Sitemap. The MVP structure; overlays sit outside the tree.
03 · FINDING THE INTERACTION
How "this matters" became a question.
Three sizes of "this matters".
Every student underlines differently and marks different things. So I made a scale: small, underline or circle with the pen; bigger, for whole sentences or lines, the question marker; largest, a dashed box that turns a whole block into one question. You use one tool all the time, or switch as you go.
Pen underlines ran into technical problems in the prototype. They were solved in the real app, where the recognisers were tested on strokes recorded from my iPad: 22 of 22, 9 of 9 and 7 of 7 marks found on three pages. One writer, a few pages: a good sign, not statistics.
A highlight isn't always a question.
In real lectures, many students highlighted the topic title just to see where the next topic starts. I do the same. So I kept the question marker and added a real highlighter, in a separate group of the tool column, so it's clear which tool does what. The toolbar took more rounds than anything else in the app.
Lo-fi. A text toolbar; the highlighter is only visual.
Hi-fi. Icons on top, paper in one tap; still one highlighter.
Shipped. A side column; the question tools are their own group.
An experiment in directing
On my earlier projects I designed everything by hand in Figma, then gave it to Claude as the reference to follow. This time I asked: what if I only direct? Most design decisions here were made through words. I described what I wanted, Claude built it, I reviewed and corrected. I went back to designing by hand only for the landing page.
I decided and directed: The vision, every decision, every review, all tests on the iPad.
Claude produced: Research drafts, the Figma screens and design system, the code of the prototype and the app.
I designed by hand: The landing page pass in Figma.
The first hero zoomed into a screenshot: very pixelated. I asked for a blank sheet that writes itself instead; the handwriting now draws from the real vector strokes.
The drawn iPad was ugly, to put it mildly. I insisted on Apple's real product bezels.
Illustrations ran into the text. The page had more lines than my Figma: the text blocks were wider than designed.
04 · THE PRODUCT IN MOTION
Mark it. It comes back. Right before the exam.
The clickable web prototype with demo data, which I tested on my iPad before building the native app. Nothing is saved. Open the prototype
One decision per screen
Native app, iPad simulator, demo content.
Page. The marker, the dashed box and the pen all make questions; a Q tab on the edge shows what's in. (shown above).
Quiz. Your own page with the answer covered, and room to write before you reveal.
Test myself. Covers every question on the page at once.
Today. Next class, questions due today and the exam countdown on one screen.
Week. Set your classes once; move one week or every week.
PDF beside the page. Slides next to your notes; cut any part of a slide into a question.
05 · TESTING AND ITERATION
Four kinds of checks, kept apart.
They are worth different things, so I never mixed them.
My own: Six rounds of testing the web prototype on my iPad, then every native build.
Real people: Friends used the web prototype; three TestFlight testers sent feedback, two of them after writing real lectures with it. No counts, no recordings.
Recorded strokes: Pen recognisers tested on real handwriting from my iPad.
AI-simulated: 8 simulated testers: an early, cheap check, not user data.
Four changes that came from testing
Change 1 · Guessing gave way to a tool
Source: My own
Before: Questions came from pen underlines and circles. At lecture speed it worked about 1 time in 4.
After: The marker always makes a question, any messy shape; the pen stays a bonus.
Why: Predictable beats clever when you're writing fast.
Change 2 · Boxing a worked example deleted its steps
Source: My own
Before: A bigger mark replaced the questions inside it.
After: Nested questions are kept.
Why: I reversed my own rule the same day, after a maths test.
Change 3 · "Undo only remembers 7 moves"
Source: TestFlight tester
Before: The history reset every time you left a notebook.
After: 100 steps, kept when you come back.
Why: The complaint pointed at the cause, not the limit.
Change 4 · "Let me bring my Notability notebooks"
Source: TestFlight tester
Before: No way in from another app.
After: Export as PDF there, import into a new notebook here.
Why: Switching cost was the reason to stay.
06 · REFLECTION
Directing from above.
What worked
The tracking system: a calendar of classes and exams that knows what's next, and the quiz statistics. Testers singled them out.
What I learned
I wanted to try being a visionary who runs the process from above and steps in by hand when it's needed. That role worked: I kept the vision of the whole project, from inside and from outside. It's a skill I'll take into my next projects.
What I'd do differently
Many of my edits went into the toolbars. Next time I'd sit down first and study how toolbars work and what makes them comfortable, and get there faster, in fewer tries.
0 likes · 0 views · Nov 5, 2026
iPad notebook that turns marked notes into practice questions and schedules reviews to improve retention; native iPad app and landing page in TestFlight.
