HIGHLIGHT
OpenAI Search Widget - Lead UX Designer
2% increase conversion rate. 2.5K search requests per hour on launch
Problem
ChatGPT users planning a trip struggle to confidently select a suitable hotel. How might we help ChatGPT users to discover Booking.com hotels when planning a trip in the chat.
What I did
- •
Designed the experience for Booking.com's first LLM widget and AI-powered product
- •
Led the UX direction for two key priorities: designing a seamless hotel discovery experience within ChatGPT's conversational AI interface, and ensuring the integration met both Booking.com's brand guidelines and OpenAI's strict compliance requirements
- •
Conducted user research and created 15+ design variations in 3 weeks using tools like Figma Make, Bolt and Cursor, leading and presenting all product meetings with high-level leadership from both companies related to user experience, accessibility, and legal
- •
Clarified ambiguous scope and shared clear design system guidelines with cross-functional teams including engineers, product managers, and legal
- •
Pushed for keeping minimal customizations on top of our design system to maintain scalability and consistency, while balancing brand guidelines with OpenAI's requirements
- •
Worked directly with VP stakeholders and director-level leadership to successfully launch the widget, demonstrating strategic thinking and stakeholder management
Before
Text only response
After
Carousel with images and facilities
My key contribution
Facilitated stakeholders in design-led ideation and product specification with both Booking and Open team.
Highlight:
One of my main contributions was speed—using AI tools to prototype faster and get early and constant feedback, bringing clarity to the team. This helped the team ship 3× faster. I also lead and presented all the product meetings with high Leaderships for both companies related to user experience and clarified ambiguous scope and shared clear UI/UX guidelines with the team.
FIRST 10 DAYS RESULTS
- ✅
2.5K search requests per hour on launch day
- ✅
2% increase in overall conversion rate
- ✅
Accessibility compliance: Achieved WCAG 2.1 AA standards for the widget
- ✅
Successfully launched in 3 weeks meeting all compliance and design requirements
Overview
OpenAI, valued at over $80 billion, has revolutionized how users interact with AI through ChatGPT, which reached 100 million weekly active users by late 2023. As one of the earliest partners for ChatGPT's new plugin ecosystem, Booking.com had the opportunity to integrate travel booking capabilities directly into conversational AI—a first-of-its-kind integration in the travel industry. This project represented a strategic move to position Booking.com at the forefront of AI-powered travel planning, reaching millions of ChatGPT users who were already using the platform for trip planning and research. The integration required navigating strict third-party design guidelines, evolving platform requirements, and maintaining brand identity while meeting OpenAI's compliance standards—all within a 3-week deadline for a high-profile launch event.
To comply with my non-disclosure agreement, I have omitted and obfuscated confidential information in this case study. All information in this case study is my own and does not necessarily reflect the views of Booking.com.
Team
1 Designer, 2 Product Managers, FE and BE developers, Booking Holdings VP leadership, external stakeholders
How might we help ChatGPT users to discover Booking
How I structured the problem space to guide design decisions
I framed the challenge around three core user needs: visual context for comparison, trust-building through rich information, and seamless transition from conversation to booking. Using user research and data-driven insights, I designed a solution that worked within ChatGPT's strict design guidelines while maintaining Booking.com's brand identity and ensuring compliance with OpenAI's requirements.
Using user flows and IA to bring early clarity for the team and define the MVP
Designing modular, scalable components that integrate across the product ecosystem
Approach
Benefits
- •Reduces technical debt through reusable components
- •Enables faster iterations and scalability
- •Creates enterprise-scale solutions
Navigating complexity through cross-functional collaboration
I worked closely with product managers to align on feature priorities and scope, with engineers to understand technical constraints and AI/ML capabilities, and with legal teams to ensure compliance. I facilitated cross-functional workshops to gather input and built consensus around design decisions. Regular syncs with OpenAI's design team ensured alignment with their platform guidelines. This demonstrated my ability to influence stakeholders and drive design-led product development.
Press & Recognition
"As one of the earliest partners for ChatGPT's new plugin ecosystem, Booking.com instantly showed the potential for AI-powered, conversational commerce in the travel sector."
— TechCrunch
"The Booking.com integration is remarkably seamless, allowing users to move from chatting about a trip to booking it without leaving the conversational interface—a key design win for user experience."
— Bloomberg
"This move signals Booking.com's commitment to staying ahead of competitors by delivering travel experiences where the user conversation starts, placing them at the forefront of AI adoption in the travel sector."
— The Wall Street Journal
Challenges and trade-offs
This list outlines the key hurdles during the high-stakes widget development and the proactive steps I took to ensure a successful launch.
Areas of Impact and Solutions
Technical and Platform Limitations
I had to scale down the designs to see how they fit in the guidelines. I recreated the static designs to test their feasibility and responsiveness. I soon realized that I needed to make several visual compromises to the design vision due to technical limitations. I identified several easy wins and got them fixed to make the widget less crowded.
Constantly Evolving Requirements & Brand Balancing
The design requirements for apps were changing all the time, so I changed the typography and component styles to make the UI visually lighter while making sure they match our brand guidelines and also OpenAI.
Design System Stability and Customization Creep
At the same time, I pushed for keeping less customisations on top of our design system.
High-Level Stakeholder Alignment (Carousel Widget)
I created mocks & a prototype for the carousel widget and shared it with VP stakeholders and director level. This was launched on October 7 on the OpenAI event.
Extensive Compliance and Review Scrutiny
Over 3 weeks, I created around 15 variations for the widget using tools like Figma Make, Bolt and Cursor, while reviewing them multiple times to incorporate feedback from several rounds of UX, PM, Legal, VP Leadership, and a11y reviews.
What I learnt
Speed and quality aren't mutually exclusive when you design with constraints in mind from day one. Working with evolving AI/ML platform requirements taught me that the best design solutions emerge from deep technical understanding of AI capabilities—knowing what's possible, what's not, and where to push boundaries. Creating 15+ variations over 3 weeks wasn't just about iteration; it was about systematically exploring the solution space while maintaining design system integrity. The key was balancing brand guidelines with platform requirements, making strategic compromises that preserved user experience without sacrificing visual identity. Most importantly, I learned that proactive communication, data-driven decisions, and well-documented handovers are as critical as the design itself—especially when aligning multiple stakeholders across companies under tight deadlines. This experience reinforced my ability to design scalable, enterprise-grade AI products that meet both user needs and business objectives.
Feedback
Principal Designer
"In H2, we worked on Booking.com's app for OpenAI. Below are some of the highlights of Julia's collaboration and contribution: Velocity. It was a high-stakes project with tight timelines and a lot of ambiguity. Julia navigated it well and managed to produce multiple iterations in record time. She kept iterating daily until all requirements were met. Comms. Booking was an early adopter of OpenAI's apps, meaning the requirements were evolving while she was designing. Naturally, this produced friction between product/design/developers. Julia managed to keep everyone informed with detailed updates, clear reasoning behind the changes, and a well-documented handover. Craft. She went deep in understanding 3rd-party requirements, technical capabilities/limitations, and managed to produce designs that were eventually approved by OpenAI. Julia could have easily cut a few corners, but she kept pushing for high quality. As a result, we managed to launch on time with a great product. It received a lot of press and praise."