Shani Leibner

Mobile, wellbeing

Dream Journal

Ever woken up from a dream you couldn't stop thinking about?

Dream Insights: dreams recorded, day streak and recurring dream patterns
Record screen: Tap to record the dream, with a glowing orb
Dream calendar where each night shows its dream's generated image

Every dream leaves a message. Dream Journal helps you read it.
Capture it the moment you wake, and watch its symbols, emotions and patterns come into focus.

A bilingual app that captures a dream before it's forgotten, finds the patterns that repeat across your nights, and gives each dream a visual, meaningful interpretation.

Role
Solo: research, UX, UI, build
Platform
Mobile, Hebrew and English
Type
Self initiated
Year
2025–26

The problem

Why do I keep dreaming this?
Most people have asked it. Dreams repeat their symbols, but the feeling fades within minutes, so the pattern never gets noticed.
The app catches the dream before it's forgotten, and shows what keeps coming back.

Who it's for

For anyone who wants a personal dream journal: a private place to record every dream and slowly discover the patterns in their sleep.

Research

The research took several directions, each answering a different part of the question: What can visual dream imagery tell us, and how can we turn it into something people can explore?

  1. 01

    The cultural lens

    I explored visual imagery in contemporary Israeli dreams, looking for recurring themes, emotions, colors, and situations. This included academic research alongside a collection of anonymized dream accounts.

    • Themes
    • Emotions
    • Colors
    • Situations
  2. 02

    The symbolic lens

    I looked at how dreams have been interpreted across psychological and semiotic frameworks, including Jungian archetypes and the relationship between symbols, emotions, and personal meaning.

    • Jungian archetypes
    • Symbols
    • Personal meaning
  3. 03

    The personal lens

    I examined how the same image can carry completely different meanings depending on the dreamer's experiences and associations. This became an important constraint: the app could suggest interpretations, but couldn't treat them as universal truths.

    • Experiences
    • Associations
  4. 04

    Turning research into a system

    The different research paths were translated into a framework for reading dreams consistently, considering their imagery, emotional intensity, recurring elements, and the dreamer's own context.

    • Imagery
    • Emotional intensity
    • Recurring elements
    • Context

All of these layers were tested against real, anonymized dream accounts and became the foundation for the product's visual language, interpretation model, and core features.

UX decisions

Three calls shaped how the app feels to use.

01

A calendar that's also a gallery

Tension
A dream journal needs a timeline, but a list of entries is easy to forget.
Call
Each night on the calendar shows its dream's generated image.
Why
The calendar holds the timeline. The images make it worth revisiting.
02

From single dreams to recurring patterns

Tension
One dream explains a moment, not why the same themes keep returning.
Call
Analyze dreams across months to surface recurring symbols and themes.
Why
Only the long view can answer "Why do I keep dreaming this?"
03

Voice first, writing one tap away

Tension
Research showed speaking captures a dream faster, but some mornings writing feels right.
Call
"Tap to record" as the main action, with "Write instead" always one tap away.
Why
Less effort to catch a fading memory, without taking away the choice.

The concept: a loop, not a log

  1. 01CaptureVoice or text
  2. 02Break downSymbols, emotion, association
  3. 03InterpretGeneral and personal
  4. 04See itA generated image
  5. 05TrackPatterns over time

The tracking step is what separates it from a journal: recurring symbols and emotions surface on their own.

The design: nocturnal, soft, readable at 6 a.m.

A night sky system of navy and violet gradients with frosted glass surfaces. Emotions are tagged in color so patterns are visible at a glance. The app is right to left first, with Hebrew as the primary language, and the onboarding copy is written to feel calm rather than clinical.

Record screen: I'm listening to your dream, with a glowing orb
Capture. Voice first, because the dream fades while you type.
Dream calendar where each night shows its dream's generated image
Breakdown. Symbols and emotions, each with a general and a personal reading.
A single dream, Sweet Cotton Fields, with its generated image and the What does it say? reading
Gallery. Generated images make months of dreams scannable.
Dream app design system: Urbanist type scale, color tokens from Midnight to Nebula cyan, navigation, action and filter states, glass buttons and the icon grid
Design tokens. One source of truth for Figma and code.

Key screens

Swipe through the screens. Tap the one in the middle to flip it and read what it does.

  • OnboardingCapture your dream while the details are still fresh.

  • Record and loaderSpeak freely while Lucid captures the details of your dream.

  • Date galleryExplore the dreams you captured on a single night.

  • Individual dreamA visual interpretation of the dream, shaped by its symbols, emotions, and details.

  • Category gallerySee recurring symbols and themes across your dreams.

1 / 5

Building it myself

This was the first app I built on my own, and the first time I connected it to APIs. I designed it, then took it all the way to working code.

  1. Research & design01Research question
  2. 02Concept
  3. 03Design
  4. 04Design system
  5. 05Wireframes
  6. Build06Screens in codeTurning the Figma screens into real layouts is where CSS finally clicked for me.
  7. 07Image generation APIMy first API connection. I chose which image service to connect, then taught it how to turn a dream into an image, so every dream gets its own.
  8. 08Interpretation botClaude reads the dream and returns symbols, emotions and meaning as structured data.
  9. 09Voice recordingSpeaking the dream instead of typing it, wired into the same flow.
  10. 10Final resultA working app, with every piece connected.

A full design token system and a written prompt guide let me move designs into code with exact values, not approximations. The analysis engine is a TypeScript service that uses Claude through the Anthropic SDK, with tool use to return structured output: symbols, emotions and interpretation as data the UI can render.

What building it taught me

Designing for AI from the inside changed how I design for it. Structure the output, and the interface can be calm.

  1. I picked up tools I had never used and learned them by building.
  2. When something didn't work, I found the way through on my own.
  3. Screens, APIs, data and voice only become an app once every piece is wired together.
  4. Building the layouts myself showed me how the design actually becomes a page.

Key Learnings & Future Direction

Key Learnings

  1. 01
    Start with the build.

    I designed first and learned the tools along the way. Today I'd start by learning how to build it and setting up the image and video generation, then move to design.

  2. 02
    A design system before code.

    Every design change during implementation slowed me down. A system set up front would have saved that.

  3. 03
    Test every day, not once.

    About 7 people used the app daily, and I improved it every day based on how they used it.

Future Direction

  1. 01
    Video instead of an image.

    Users asked to see their dream as a video.

  2. 02
    Interpretation length.

    Some wanted it shorter, others wanted more room for it. Next step: test both.