Back to home

Back to home

Bubbly AI App

Multi agents travel assistant app

Overview

Solution

Key Features

context

Iterations

Linked In

Resume

Email

Product designer who finds joy in scaling products with on-brand craft and intention

What if AI agents pick our travel destinations?

Bubbly AI is an AI concept design project, focusing on quick design iterations on swift UI, creating a fun and delightful engagement with AI agent based decision making.

2025 PRODUCT DESIGN

AI Concept

UI design

IOS

BUBBLY AI APP

Multi agents travel asistant app

Overview

CLIENT

Self

Time

2025

Team

1 Dev, 1 Designer (Me)

Role

Product Designer

Tools

Swift. Figma

My Role

Curious about rumors of Apple’s emerging glass-inspired design language, I teamed up with a developer on a self-driven iOS side project to experiment, learn, and turn playful interaction ideas into a working product.

Curious about rumors of Apple’s emerging glass-inspired design language, I teamed up with a developer on a self-driven iOS side project to experiment, learn, and turn playful interaction ideas into a working product.

Solution

An AI that allows multi-persona decision making

Key Features

Feature 1

Create your proxy with bubbles

Make trip planning fun. Set preferences and weights through playful bubbles and build your AI trip proxy.

It took a lot of iterations to get the physics right, but I'm so happy with how it turned out!

Feature 2

Talk to fine-tune your trip taste

Interact with AI to adjust preference weights or receive recommended tags.

Feature 3

Put your cards on the table ♣️

AI will express your preferences for you, making it easier for the group to decide together.

Feature 4

Deal maker!

No drama, no oversharing. Bubbly will curate options based on each proxy's unique preferences analyses.

Context

Everyone is talking about AI but I found it hard to dive deeper into "UI" part for AI products.

Using the latest iOS 26 glass interfaces, I explored creating delightful interactions with "bubbles".

Problem

We have AI agents. Why not make them help group decision-making?

  1. Current AIs are built for individuals, not groups

  1. Group decisions are messy and difficult

Competitive Analysis

I tested a scenario of group trip planning with existing apps like Airbnb Tripadvisor, and Hotels.com.

Airbnb

Airbnb had the most usable & transparent voting feature to support group decision making.

Airbnb

Airbnb had the most usable & transparent voting feature to support group decision making.

Tripadvisor

Tripadvisor also had a group wishlist, but lacked additional features to share opinions beyond text-based comments.

Initial brainstorming

Through competitive analysis, I discovered design opportunities.

No AI feature for groups

Even when AI was used, it was 1 person recommendation system and it didn't work in group decision-making.

AI for a group, not just a solo

Recommender system that can interpret & reconcile inputs from multiple users, generating group recs.

No group-aware settings

None of the platforms allows let users collaboratively set filters/ preferences. There’s no way to balance or weigh.

Enable dynamic preferences

The new design should allow users to set filters collaboratively, and adjust individual priorities and preferences.

No group-aware settings

Airbnb has up/down voting, and TripAdvisor has comments. But neither helps shared decision formation,

No more manual work

With a criteria-based recommender, we can reduce the burden of manual negotiations

Concept

Bubble-inspired, playful trip planning!

Inspired by Apple’s Image Playground, I created a moodboard for an interaction that combines a glass-like aesthetic with flexible, dynamic behavior.

Wireframes

Considering components, and the overall design system, I quickly sketched the main flows.

Flow 1

Editing preferences

  1. Edit preference for my proxy

Flow 2

Group-AI curation

  1. Adjust weight of each preferences

  2. Share my proxy with friends

Flow 3

Voice and image-based AI UX

  1. Chat with AI to adjust preference weight or receive recommended tags.

  2. Upload images to get more personalized or related keywords (bubbles).

Design iterations

Detail 1

Intentional craft to reinforce the AI experience

Gradient UI

I used gradients selectively to make the interface feel more dynamic and responsive without overwhelming the core experience.

The pick-up CTA, given its highest priority, was placed in a consistent location across different screens to enhance the seamless pick-up experience.

Liquid glass aesthetic

The developer and I had been exploring glass-inspired UI from the early stages of the project. With the release of iOS 26, Liquid Glass aligned naturally with our existing direction, allowing us to integrate the new visual language seamlessly into the experience.

The pick-up CTA, given its highest priority, was placed in a consistent location across different screens to enhance the seamless pick-up experience.

Detail 2

Small tweaks to engage users

Bringing the card game metaphor into the UI

First and foremost, I wanted trip planning to feel fun, like playing a board game with friends. As each member’s preferences are highlighted, recommendations appear as cards that everyone can place and explore together, turning the stress of group decision-making into a playful, collaborative experience.

The pick-up CTA, given its highest priority, was placed in a consistent location across different screens to enhance the seamless pick-up experience.

Detail 3

Crafting playful moments

Turning tedious profile setup into play

Profile setup is often one of the most manual and tedious parts of the experience. I wanted to turn that friction into a moment of play by making preferences tangible, interactive, and fun to explore.

The pick-up CTA, given its highest priority, was placed in a consistent location across different screens to enhance the seamless pick-up experience.

Real-time physics for the bubbles

We went through multiple iterations to find the right balance of physics values, making the bubbles feel realistic yet playful. To make preference weighting more intuitive, I also designed a cycling size interaction that lets users adjust importance directly through the bubbles themselves.

The pick-up CTA, given its highest priority, was placed in a consistent location across different screens to enhance the seamless pick-up experience.

The pick-up CTA, given its highest priority, was placed in a consistent location across different screens to enhance the seamless pick-up experience.

Reflections

Ideas to implementation

This was my first time working on a UI-heavy project directly with a developer. I learned how to clearly translate UI and interaction needs into developer-friendly terms while iterating through constant collaboration.

Explore, learn, enjoy

At first, I felt pressured to build an AI concept by strictly following UI “must-dos.” But I soon realized it was restricting my creativity and approached it as a creative experiment. This mindset gave me the freedom to explore more.