Aaaaaaname undisclosed · 2026 · Product design and iOS build
Helping newcomers in Germany understand official letters
A letter in German can carry a deadline and a penalty. I designed the app so a newcomer can scan one, understand it in their own language and know exactly what to do next, and I built it as an iOS app.
0 to 1 productAI assistantiOS appTrust and privacyOnboardingAI prototyping
The three intro slides. The faded screens are the one before and the one after.
Product
Aaaaaaname undisclosed · iOS app for newcomers in Germany
Year
2026
Role
Product designer and builder
Tools
Figma, SwiftUI, Claude Code
Problem
Letters that nobody explains
Offices in Germany write in formal German, and the letters matter: a missed reply can mean a fee, a fine or a delayed permit. Newcomers often cannot tell which letters are urgent, so they ask friends, paste them into a translator or ignore them.
Hard to read: formal German, unfamiliar offices, deadlines buried in the text
Hard to trust: the letters hold names, addresses and tax numbers, so sending them to an AI feels risky
Hard to follow up: a reply, a missing document and a date live in three different places
Research
How big is the need?
Before I drew a screen, I sized the need from public figures, then looked for evidence that the letters themselves are the problem.
14.1MForeign nationals living in Germany at the end of 2025
1.69MPeople who moved to Germany in 2024, 430,000 of them net
32%Of people who emigrated named cumbersome bureaucracy as a reason
Two things stood out. A large group lands in the same system every year and gets the same formal letters. And I could not find a study that measures whether newcomers understand those letters. Caritas calls official German the biggest barrier for migrants, but that is an advocate's view, not a measurement. So the first thing the app has to prove is comprehension.
Everyone
About 14.1 million foreign nationals. I count people, not revenue, because the MVP has no payments.
Who I serve
The 1.69 million who arrived in 2024, since I assume the first year carries the heaviest paperwork. It is an upper bound: the figure includes every nationality, EU citizens and returning Germans too.
First goal
Not a market share. A test group of newcomers, large enough to learn whether the explanation works.
If a newcomer can scan an official letter and get a plain explanation, the deadline and one approved next step in their own language, they will act before the deadline, and trust the app enough to keep their letters in it.
That sentence hides several bets. These are the riskiest, with how I would test each one.
The riskiest assumptions
Value
Newcomers act sooner when a letter arrives with a plain explanation and one next step. Confidence: medium. Test: give 8 newcomers a real letter, once through the app and once their usual way, and compare the time to a correct next step.
Trust
People will scan real letters if documents stay on the phone and the redaction is visible. Confidence: low. Test: ask testers to scan one of their own letters, count who does, and listen to what they say about the privacy screen.
Usability
A first-time user finishes onboarding in under 3 minutes and reads an explanation within 15 seconds of scanning. Confidence: medium. Test: timed sessions on a real phone over 4G.
Feasibility
On-device text recognition plus redaction is accurate enough on real German letters. Confidence: medium. Test: run 30 sample letters through the redactor and count the names, addresses and numbers it misses. Any miss is a blocker.
Viability
Explaining a letter is free everywhere, so people will pay for the follow-through: reminders, drafted replies and the appointment pack. Confidence: low. Test: not in the MVP, which has no payments. After the first round, put a price on the follow-through and see who asks.
Go to market
I can reach newcomers when their first letter arrives, through language schools, integration courses and expat groups. Confidence: low. Test: pilot one channel and count how many people install and scan a letter within a week.
Ethics
A drafted reply can mislead someone. Designed against: every outgoing action goes through a preview with an approve button. Test: watch whether testers read the preview before they approve.
How I would measure it
The app is productivity software, so the measure is how much of the paperwork actually gets done.
North Star: letters handled on time, the share of scanned letters whose next step is finished before the deadline.
Scan to explanation: under 15 seconds on 4G
Onboarding: finished in under 3 minutes
Photo to approved next step: under 5 minutes
Reminders: every deadline has them at 7, 2 and 0 days
Trust signal: replies approved as drafted against replies edited or discarded, read next to the others and not a goal on its own
The app has no analytics or tracking SDK and I want to keep it that way. So the plan for the first round is moderated sessions with screen recording, plus counters stored on the device that a tester can show me. Any usage numbers after launch would be opt-in and aggregate, and would never include letter content.
I set the bar before testing: the first round is a go if at least 6 of 8 testers name the correct next step and deadline faster than with a translator, and at least 6 of 8 scan one of their own letters.
Approach
How I tackled it
Three decisions shaped the product:
Explain, then act. Every letter ends in a clear next step, not just a translation.
Trust first. Documents stay on the phone, and only redacted text is ever sent for analysis.
Home is for what to do. Statuses and actions live on Home. Documents is plain storage.
Solution
What I built
Scan and explainAI
1. Read the letter and explain it
Scan a letter and the app reads the text on the phone, finds the sender and the deadlines, then explains it in the user's language with the next steps. Every AI screen says "Explained by AI", and users can set a status by hand if they would rather not use a drafted reply.
Four reading steps, then a plain explanation with the deadline and the steps.
To-do and deadlinesClarity
2. Know what to do next
Home gathers every action into one to-do list, soonest first, with a status on each procedure. Deadlines is a calendar colored by urgency. Opening a procedure shows its deadlines and the full history of letters and replies.
One list of actions on Home, one calendar for dates. Opening an event leads to the procedure, with its deadlines and history.
PrivacyTrust UX
3. Private by default
Letters and documents are stored only on the device and are locked with Face ID. Names, addresses and numbers are replaced with placeholders before any text leaves the phone, and restored on the device. The privacy promise is the second thing a new user sees, in one plain sentence.
The privacy promise, before any permission is requested.
Process
How I built it
I worked spec first. The spec is one document written for Claude Code: the goal, success criteria, data model, privacy rules and six build phases. The privacy rules are hard requirements, so if a task seems to need a change to them, the rule is to stop and ask, not to find a workaround. The design reference is a canvas of 22 screens, and where the spec and the canvas disagree, the spec wins.
Phase 1: project setup, design system and all 22 screens with static content
Phase 2: storage, file protection, app lock and delete everything
Phase 3: the document vault, with scanning and on-device text recognition
Phase 4: the analysis service, redaction and the scan-to-explanation flow
Phase 5: reminders, calendar export and hand-off
Phase 6: an accessibility pass and UI tests for the three main flows
After each phase I stopped, ran the checks and reviewed it on the simulator before moving on. After phase six I redesigned Home, Deadlines and Documents, and added illustrations, a palette and motion.
Prototype
Try it yourself
A clickable version of the app. Open a to-do, read the letter, prepare a reply and send it, then check how Home and Deadlines change.