Four white smartphones with dreaimade screens among purple 3D shapes

dreaimade
dreams are made to create

An AI-based service that turns dreams into a source of ideas

An app that wakes you shortly after you fall asleep, records your dream by voice and uses AI to visualise it as a picture story.

My role: Solo project, from research and an expert interview through concept, interaction and UI design to prototype and case film.

HAWK Faculty Award Presented at theADC “So Future” Day
Detail of the dreaimade app
Creatives

have inspiring dreams particularly often

57 %

of nights leave no memory of a dream

70 %

of dreams pick up on a topic set beforehand

78 %

more creative after a nap with dream incubation than after no sleep

Sources: expert interview with Prof. Dr. Michael Schredl · Schredl (2013), Träume – Unser nächtliches Kopfkino · Horowitz et al. (2023), Scientific Reports

Facts

Master’s project 2024
Service Design

Supervised by:
Prof. Stefan Wölwer, Vincent Timm

Methods

Research through design · expert interview · letter to grandma · reverse brainstorming · personas · interviews · user journeys · ecosystem map · user flow chart · wireframing · low-fi & high-fi prototyping · cognitive walkthrough · usability testing · motion design

Tools

Figma · FigJam · Blender · After Effects · Adobe Firefly · ElevenLabs · ChatGPT · Krea AI · RunwayML · Genmo

Problem

The most creative thoughts come as you fall asleep, and by morning they’re forgotten

Image generated with Adobe Firefly: a student asleep on the sofa with a holographic idea bubble above him

The design students I interviewed mostly look for inspiration on Pinterest and Instagram. Coming up with original ideas is often hard.

Studies show thatdream incubation in the N1 phase, the onset of sleep, can boost creativity afterwards.

Dream journals such as Alchera or the “Träume” app only record and interpret dreams. No app uses sleep-onset dreams as a source of ideas.

First approach

What seemed obvious at first

Why dreaimade doesn’t interpret dreams

The personal context is missing Expert interview

What a dream means depends on what someone has experienced. An AI doesn’t have that knowledge.

People want inspiration, not explanations Interviews

The target group gets ideas from images and mood boards. An interpretation doesn’t help them come up with ideas.

So it only asks questions and visualises

The AI in dreaimade only asks follow-up questions and visualises the dream stories. According to Prof. Dr. Schredl, AI can help people remember dreams better through targeted questions.

Research

Sleep-onset dreams are particularly creative and link the dream to what you were thinking about beforehand, which means they could be inspiration for new ideas.

Prof. Dr. Michael Schredl, expert interview

Concept

From insight to design decision

I spoke with sleep researcher Prof. Dr. Michael Schredl and four people from the target group. The key design decisions are based on those conversations.

Interviews Open brief without a topic

In the interviews, an open brief with no starting point was the most common obstacle to coming up with ideas.

Topic before falling asleep

Before a nap, users set a topic that steers the dream.

Studies Waking left to chance

An old method: you fell asleep holding an iron ball. When it dropped, you woke up and could capture the thoughts you’d had while drifting off.

Waking in the N1 sleep phase

dreaimade detects the N1 phase via motion sensors and wakes you 1 to 5 minutes later.

Expert interview Typing with eyes open

Opening your eyes to type makes you forget the dream quickly.

Voice instead of typing

You describe the dream out loud, with your eyes still closed.

Requirements for the service

Reverse brainstorming in FigJam with the columns Collect Ideas, Find Problem and Find Solutions Reverse Brainstorming
Wake gently

A loud alarm would jolt you out of light N1 sleep, so dreaimade wakes you with a soft tone and a calm voice.

User journey map “falling asleep” with gain and pain points User Journey
Recording starts automatically

Right after waking you’re too tired to look for your phone. Recording starts as soon as the alarm goes off.

Excerpt from the user flow chart: from first launch through onboarding to recording the sleep-onset dream Flowchart
Measure the sleep phase first

Before dreaimade can wake you, the app has to detect the sleep phase, so tracking via smartphone or smartwatch is set up right after sign-up.

Iteration

Cognitive walkthrough via Zoom with Michelle Bümmerstede: low-fi prototype on the iPhone and guiding questions in FigJam

What testing changed

An onboarding flow most people skipped

First solution

In the cognitive walkthrough, Michelle Bümmerstede (UX/UI at CEWE) pointed out that, on first login, the service didn’t explain why it’s worth using. So I added an onboarding flow.

What happened in testing

Most of the four test participants skipped the “Show me dreaimade” prompt, so the problem wasn’t solved yet.

Second solution

Until the first dream has been recorded, dreaimade explains each step right where it’s needed.

UI design

Figma for the look, AI for the dream visualisations

dreaimade micro design system with colours, typography, icons, buttons, inputs and cards
Micro design system

I first defined the look with style frames, a method from motion design. Building on that, I created a micro design system in Figma with colours, typography, icons, buttons and cards.

Three dream images generated with Adobe Firefly from the “Balls of memories” visualisation
Generative AI

I simulated the dream stories in the prototype with generative AI: text with ChatGPT, images with Adobe Firefly, video and voices with tools like RunwayML and ElevenLabs.

Result

How dreaimade works

  1. Set a topic and talk about it with dreaimade
  2. Fall asleep and get woken shortly after the N1 phase begins
  3. Talk about the dream with dreaimade
  4. Fall asleep again
  5. Wake up and visualise the dream story together
  6. Use the dream story as inspiration for your own ideas
Screens of the dreaimade app

Created in Figma

Try dreaimade

View prototype

Learnings

What I’m taking away

Bring in expertise

Without the expert interview, dreaimade wouldn’t exist. A concrete idea only emerged once I had structured that knowledge.

Get fresh eyes early

After weeks on the project, a lot seemed obvious to me. The cognitive walkthrough surfaced things I’d assumed were self-explanatory, for example that the headphone icon was read as music.

Get comfortable with ambiguity

I started with nothing but the topic of dreams and no finished idea. With research through design, I alternated between researching and designing, and the idea emerged through experimentation.

Documentation

Cover of the dreaimade documentation

dreaimade documentation

The full documentation of my master’s project, with research, interviews, tests and prototype, as a PDF (in German).

View documentation

Next project

AR Service Design

iYoou: You move and I grow

An AR avatar that is only visible when you stand up and grows with every movement, nudging students to break up long periods of sitting.