Problem
40% of US travellers use AI tools for trip planning (PhocusWright), a shift that's pulling traffic away from traditional search experiences like Expedia. The problem for travellers is that AI recommendations are generic, because frontier LLMs don't have the specific, local knowledge needed to give quality answers. Creators have that knowledge, but their audiences send so many trip planning requests that responding individually doesn't scale alongside running a content business.
Task
If EG gave creators AI agent tools to handle trip planning at scale, creators would choose EG as their platform, bringing their audiences with them for discovery and booking. I designed an AI-powered agent experience to test this, where creators could respond to trip planning requests from their audience with personalised recommendations, while surfacing EG's inventory as the answer.
The Creator Experience
Creators needed a way to respond to every trip planning request without it consuming the time they spent running their business. AI agents can handle complex, specific requests autonomously when given the right context, which made them a strong fit for this problem.
I designed a system where each creator trained their own agent by providing instructions and past recommendations from their blogs, social platforms, and EG's creator tools like Travel Shops. The agent learned their recommendations and tone of voice to respond to travellers as if it were the creator. Every hotel and activity recommendation was tagged with an affiliate link through our creator platform, so creators earned a commission on bookings.
The Traveller Experience
Travellers wanted recommendations as personalised as asking a creator directly. When trained on a creator's knowledge and voice, the agent could deliver that experience at scale.
The chat experience let travellers discover and plan an entire trip through the agent. They could request recommendations based on their needs and the needs of everyone in their group. They could also request detailed itineraries that accounted for individual schedules and constraints — for example, if someone had already booked something and needed to plan around a specific destination and date.
Achievements
I was recognised for leading the design and engineering of the experience, earning the AI MVP Award for 2025. It was one of the more rewarding projects I've worked on, designing a product where the incentives genuinely aligned for EG, creators, and travellers at the same time.