I defined the IA and interaction model for KAYAK’s AI-driven travel search, creating the foundation that connects LLM prompts to structured metasearch results.
The project began in June 2024 with a small four-person team exploring how natural-language queries could translate into actionable travel searches.
This groundwork later enabled the product to scale and became a core part of KAYAK’s broader AI roadmap.
How KAYAK.ai Works
Designing the AI Results Page
I focused on how to present KAYAK’s live travel data in a way that clearly differentiates KAYAK.ai from other LLM tools.
Rather than emphasizing visual detail first, I designed for reusability and scalability, exploring how GPT-generated responses and KAYAK’s structured data could be combined into a unified, adaptable results layout.
Final Design: Phase 1
The first launched version covers the full user scenario — from getting a sense check on potential destinations, to finding the best time to travel, to comparing flights along a timeline, and finally booking a flight on KAYAK.
Final Design: Key screens from the Explore → Flights flow
Phase 2 Design Goal: Expanding AI Components
After the first launch, we needed new features quickly to broaden what KAYAK.ai could support.
The challenge was finding a flexible way to scale an early-stage product with many moving variables. To address this, I created a compact AI design set based on KAYAK’s existing design system and built an atomic UI structure that could be assembled and expanded easily.
System-Driven Design Approach: Light Theming
This system-driven approach also helped us respond to changing requirements and collaborate more efficiently across teams. Because the components were already structured for flexible adaptation, we were able to apply the new light theme smoothly with the branding team’s new brand strategy.
System-Driven Design Approach: Widget Expansion
A key area where I applied this system-first approach was the set of widgets that display KAYAK’s live travel data.
Each widget was structured with a clear header and body, allowing the two areas to serve different roles and scale consistently across features.
The unified widget structure I designed enabled rapid expansion, allowing us to scale the system into ten variations in a short period of time.
Phase 2 Improvement: All-in-One Experience in KAYAK.ai
Phase 2 also enabled users to complete detailed comparisons without leaving KAYAK.ai.
The right-side detail view reused KAYAK’s existing product pages, allowing a seamless transition while keeping the entire experience within the AI workflow.
What’s Next: Bringing AI into KAYAK.com
One year after the initial launch, a new team was formed to bring AI capabilities into KAYAK’s core product. Alongside leading the design for KAYAK.ai, I was tasked with helping transition these AI experiences into KAYAK.com.
My role now focuses on refining the early design patterns into reusable components that can be applied across any KAYAK surface.
