Atlas

Overview

Atlas reimagines travel planning by combining large language models with structured travel data. Users describe their ideal trip in natural language, and Atlas generates a detailed, bookable itinerary — complete with lodging, activities, dining, and transit.

The Challenge

Traditional travel planning involves hours of research across dozens of tabs. Existing tools either offered generic package deals or required manual assembly of every detail. We saw an opportunity to bridge the gap between inspiration and execution.

Approach

I designed the end-to-end experience from initial concept through launch, working closely with the machine learning team to define the interaction model between user intent and AI generation.

Design highlights

• Conversational input — A natural language prompt replaces the traditional multi-step form. Users describe what they want in their own words.
• Editable itinerary cards — Each generated block can be swapped, rearranged, or refined. The AI adjusts surrounding logistics automatically.
• Progressive disclosure — The interface reveals complexity only when needed, keeping the default experience clean and approachable.

Results

Atlas launched to a private beta of five thousand users. Engagement metrics showed an average session length of twelve minutes and a completion rate of seventy-three percent — significantly higher than comparable tools. The product secured seed funding within three months of its public launch.