AI-powered local discovery tool

An AI-native discovery platform that guides users from intent to action by combining personalized recommendations, map context, and user-generated content.
AI-Driven Discovery
Interactive Map
Community (UGC)
My Role:
Product Designer
Status:
Launched on web and iOS
Timeline:
March  - August 2025
Team:
1 Product designer, 2 Visual designers, 1 Product manager, 2 AI Engineers, 5 Developers
Deliverables:
  • Revamped 80% of web and iOS interfaces, enhancing usability and consistency for 2M+ users.
  • Built a design system from scratch and established design guidelines to scale design quality.
  • Conducted user research to guide decisions.

Overview

About:
Wanderboat is an AI-powered local discovery platform that transforms how people find places to eat, explore, and stay. Unlike traditional search or generic AI summaries, it blends chat-based feeds with an immersive map, surfacing curated dining, events, attractions, hotels, and experiences — all grounded in real photos and videos from real people.

Impact

22%
in project-led user growth
27%
in user engagement with AI interactions and feature exploration.
15%
in average time spent on key pages

Problem

01: Despite the product’s potential, users struggled to understand what Wanderboat could do for them among so many AI tools.
02: At a foundational level, the product wasn’t built to scale, and the absence of a unified design system led to significant UX and UI inconsistencies across platforms.
03: Engagement was low, especially on the chat page — users didn’t spend much time exploring.
Old Design

How AI Works Here

Wanderboat’s AI analyzes millions of social posts and videos in real time, using a POI engine, Thinking Agent, and Vision Understanding to deliver personalized, trustworthy local recommendations.

User Study

Based on responses from 300 users, here are the main takeaways..
AI’s role was unclear on the homepage
The homepage mixed prompts, recommendations, and auto-scrolling content in one feed, making it unclear what the AI was meant to help users do or where to begin.
Text-heavy AI responses reduced engagement
Chat responses were long and text-heavy, causing users to read them like answers rather than explore options. Users wanted clearer rationale and visual proof.
Discovery, chat, and maps felt disconnected
Discovery, chat, and maps worked as separate surfaces, making it hard for users to compare options or make quick decisions.

Design Goals

01: Clarify Product Value Through a More Guided, Intent-Driven Experience
Help users quickly understand what Wanderboat can do for them by designing clearer onboarding, contextual prompts, and guided AI interactions that highlight the product’s unique value among other AI tools.
02: Increase Engagement Through More Interactive and Rewarding Exploration
Enhance the chat and discovery flows to encourage deeper interaction. Introduce more intuitive navigation, richer content formats, and proactive AI suggestions that keep users engaged longer and motivate them to explore.
03: Establish a Scalable, Cross-Platform Design System
Create a unified design system that ensures consistent UI components, interaction patterns, and visual language across web and mobile. This foundation supports long-term scalability and supports future surfaces like events, hotels, social content, and trip planning.

Selected drafts

I worked with engineers to validate feasibility and backend data support, and with AI experts to refine chat interactions, ensuring designs were both user-friendly and technically achievable.

Ideas to action

More Guided, Intent-Driven Experience

" The homepage mixed prompts, recommendations, and auto-scrolling content in one feed, making it unclear what the AI was meant to help users do or where to begin."

Solution:

‍
1. Enhanced decision-making by connecting AI suggestions, cards, and map context, enabling users to evaluate proximity and relevance more efficiently.
‍
2. Enhanced engagement through visually rich suggested prompts that guided users toward exploration and deeper interaction with the AI chat.
‍
3. Personalized recommendations by making prompts dynamic—tailored to user location, weather, and user interests selected during onboarding.
‍
4. Introduced visual hierarchy by designing varying box sizes to add visual variety while scrolling, reserving larger boxes for more popular or important prompts.

Build trust through context-rich results

Users found AI results unclear and text-heavy, rarely engaging with recommendations, and wished for more map-based information on web and mobile.

Solution:

1. Replace long text summaries with short, scannable descriptions paired with images or video snippets.

2. The new menu tabs (Restaurants / Attractions / Events) introduce a clear information architecture that supports non-linear exploration.

3. In addition to AI recommendations, both the map view and list view display suggested place, and users can quickly switch between them using the “Map & Moments” toggle.

3. User posts and short videos are displayed under the list view, offering authentic reviews that help users make more confident decisions, without needing to open multiple tabs or search across different platforms.

user feedback

Next Project
HaveYouMet
Product designer
UX DESIGNER
Visual Designer
Product designer
UX DESIGNER
Visual Designer
Product designer
UX DESIGNER
Visual Designer
qingyu068@gmail.com

 QINGYU © 2025