Refining predictive capabilities significantly enhanced user engagement.
40
Increase in User Engagement
%
Improved alignment with client preferences through enhanced AI accuracy.
30
%
Boost in AI Accuracy
Through 3 months of beta testing and iterative development, Stellar has achieved:
PROJECT OVERVIEW
Stellar is developed by a seed-stage startup focused on solving communication gaps between designers and clients in residential architectural projects. As the Product Designer, I led the UX strategy and development of key features to address the root pain point in the architectural lifecycle—effectively capturing and translating client preferences into actionable designs.
Product Development
Stellar is being developed in phases, with each phase focusing on refining critical features based on feedback. We are currently developing Phase 1 in manageable increments.
Key Features (Phase 1):
1.1 Client Preference Decoding: Translates client preferences into editable conceptual renders, making it easier for designers to align with client expectations.
1.2 Flexible Conceptual Renders: Provides designers with three modes of conceptual rendering, allowing for more adaptive and versatile design options.
1.3 Design Iteration: Simplifies the creation of multiple design options by using editable components that can be easily swapped and modified.
Upcoming Features (Future Phases):
-
Streamlined Material Sourcing: A material repository with cost insights to help designers make informed decisions about material selection.
-
Advanced Visualization Techniques: Immersive features like VR and design personas will allow clients to visualize themselves in the spaces they’ve envisioned, enhancing the design experience.
-
Specialized Designer Repository: A repository categorizing designers by their areas of expertise, fostering easier client-designer connections and collaboration within the design community.
A comprehensive breakdown of Phase 1 detailing the process from initial client consultation to the first render:
CASE STUDY [ III ]
AI-Powered Design Assist Tool
Stellar: A PropTech Startup
An AI-powered design assist tool that helps designers capture and translate client preferences into real-time conceptual renders, offering comprehensive support throughout the project lifecycle to streamline communication and reduce costly revisions.
Role
Employer
Platform
Area
UX Designer Architectural Expert
Stellar
Web
Conceptualization, Design and Research
The communication gap between designers and clients triggers a series of issues in the design process, leading to miscommunication, delays, and frequent revisions. Current tools either focus on isolated tasks or are client-centric, but what’s needed is a comprehensive tool that supports designers from capturing client preferences to streamlining the entire workflow.
PROBLEM STATEMENT
2. Intuitive Swipe Interaction for Preference Tagging:
-
Users express their design preferences by swiping left to dislike or right to like an interior design style, making the process familiar and engaging.
-
As users swipe, the AI ranks their preferences based on the selected styles, leveraging tags and keywords to better understand their unique taste.
-
Based on user preferences, the AI will provide three personalized conceptual render options for review in the next phase of development.
Solution for Part 1.1 'Decoding Client Preferences'
1. Streamlined Client Onboarding and Data Management:
-
The dashboard consolidates all client information, allowing designers to easily track progress and manage multiple projects without switching between tools.
-
Key debriefing notes and client preferences are stored directly in the client profile, ensuring designers can align the design process with the client’s unique needs from the start.
-
The "Add New Client" feature simplifies data input, reducing time spent on administrative tasks and ensuring no critical details are missed during the onboarding process.
MVP
Client Dashboard
MVP
Game Screen with Swipe Interaction
Image Source: Author
Phase 1 Breakdown Illustration
2.0 Intuitive Map Feature
Locating the Bathhouse
The development of Stellar was based on extensive research, leveraging global insights and a user-centric approach to ensure the tool addresses industry pain points. The research process involved:
User Interviews and Global Insights: I led a user research campaign, engaging over 20 industry experts across Dubai, Egypt, the UK, the USA, and India. These experts provided diverse perspectives on the inefficiencies in client-designer collaboration. Their feedback underscored the need to streamline early design stages and reduce miscommunication.
Research Strategy
Design Collaboration
I collaborated with a product manager, two software engineers, and our board of industry experts to develop Stellar. We followed a Scrumban approach, with regular stand-ups and bi-daily meetups to keep the team aligned. Input from our board of experts was crucial throughout the process, ensuring the tool’s features addressed real-world industry needs.
Video Credit: Author
Excerpt from Interview with the Postgraduate Program Lead, School of Architecture, Oxford Brookes University
Most of the Research regarding the Personal Care Industry lead me down the same path: "Technology was seeping into the future of this Industry". However, the statistics were astounding - reflecting a changing consumer pattern that highly valued a venue's online presence when booking their services.
In a survey conducted by Lisa Hedges - an associate principal analyst at GetApp - of over 200 North American consumers, the key findings indicated that:
The Digitalization of Personal Care Services in 2023
WHITE PAPER RESEARCH
Personalization Gap in Wellness Industry Apps
COMPETITIVE ANALYSIS
Friction Point Analysis Across Project Stages: We categorized pain points across each stage of the residential project lifecycle, from initial consultation to project completion, using three comprehensive data sets based on user interviews and research. This ensured Stellar targeted the most significant challenges, directly addressing real user needs and improving the client-designer collaboration process.
Video Credit: Author
Screen Recording of Document Outlining Pain Points at Each Stage of the Project Lifecycle
Video Credit: Author
Tagging Exercise With Industry Expert
Five-Second Testing for AI Tagging Keywords: We conducted rapid tests to capture immediate reactions to popular interior styles and understand the keywords professionals used to describe them. These insights were key in training Stellar’s AI model for accurate image tagging and client preference matching.
Image Credit: Author
Conceptual Renders by Expert for AI Render Replication #1
Image Credit: Author
Conceptual Renders by Expert for AI Render Replication #2
Concept Testing: During development, we conducted multiple concept validation sessions. In one pivotal session, a team member reviewed a deck of interior styles and pinpointed preferences and dislikes. I then generated two conceptual renders, which helped us map the process end-to-end and inform the AI model.
Iterative Development: Using an Agile Scrumban approach, we continuously refined Stellar based on real-world feedback. This iterative process allowed us to adjust features to meet market needs, enhancing the tool’s predictive capabilities while keeping the design process simple.
Design PROCESS PHASE 1.1
The design process for the Phase 1.1 MVP was driven by extensive research, global insights, and user feedback, ensuring the interface was intuitive and user-friendly. I led key design decisions to create an engaging interface that aligned with user needs. Usability testing helped refine the design, allowing the AI to accurately capture client preferences and generate relevant results, ultimately improving the overall user experience.
LANDING PAGE
Option 1
A more visual-led approach
Option 2
More Text Dominated
After careful consideration, Option 2 was chosen for its clean and simple layout, which made the main message stand out clearly. We further refined the design using Base Web, enhancing its functionality and visual consistency. The displayed metrics added credibility, while Option 1's darker design reduced readability and user comfort.
Option 1.1
Refined Landing Page
Sign-Up and Login Page
Sign-Up Page
Login Page
We carried Option 1's aesthetic seamlessly through to the sign-up and login pages, ensuring visual consistency across key touchpoints. The image I originally created for Option 1 was subtly modified to evoke a sense of progression, with the open door symbolizing the user's movement into the next phase of their journey.
Client Dashboard: Access to Swiping Interface Game
First look at Client Dashboard before Game
After Completing Game and recording Client Preference
For Phase 1.1, we focused on designing a specific section of the client dashboard that integrates seamlessly with the swiping interface game.
Swiping Interface Game
Option 2
Swiping Interface
Option 1
Swiping Interface
We selected Option 2 due to its cleaner layout and enhanced user focus. The design more effectively highlights the images, making the decision-making process (swiping left or right) more intuitive for users. This streamlined experience contributes to a smoother, more engaging interaction compared to Option 1.
The Result Page presents a comprehensive breakdown of the client's design preferences, visually represented through a curated mood board and a detailed color palette. The document will be stored in the client dashboard, allowing for easy reference throughout the project, ensuring alignment between the client’s vision and the final design outcomes.
RESULT PAGE
Game Result
To be a downloadable Format for Client
