Decolar - AI Travel Agent
Role
Solo Product Designer
Timeline
1 Month (MVP-Paced)
Platform
WhatsApp Agent (Mobile)
Tools
Figma, Miro, Copilot Studio
The Context
In a strategic partnership between Elo (one of Brazil's largest card payment brands) and Decolar (a leading Latin American travel agency), this project explored how booking flights could be transformed. We designed a conversational AI Travel Assistant within WhatsApp to compress the exhaustive flight search, benefit checking, and payment checkouts into a unified experience.
The Goal: My task was to turn a broad concept into a clear, testable, and highly functional WhatsApp MVP experience in 30 days. The objective was to eliminate the traditional 5-hour travel booking journey, designing a conversation-led agent that guides users from flight exploration to checkout under 10 minutes.
The Challenge
Exhaustive Travel Booking Funnel
Buying flight tickets is historically long and stressful. According to a internal research, users spend around 5 hours checking dates, comparing flight options, applying promo codes, and validating credit card benefits across disjointed screens.
Rigid Conversational Spaces
Designing within WhatsApp is structurally limited. We had to represent rich, multi-variable flight comparisons and complex checkout transactions in text messages without generating user fatigue.
Parallel Production Under 30 Days
The functional product had to be fully developed and shipped within an aggressive 30-day window. Joining the project with zero prior design specifications, I had to build the entire visual and conversational journey from scratch in lockstep with the engineering team as they coded.
The Solution
Comparative Service Blueprinting
Mapped potential AI conversational journeys against traditional OTAs to identify friction points and determine where automated intelligence adds most value.
Synthetic Persona Testing via Own AI Agent
Utilized a user-simulation AI agent that I previously developed myself to speed up design validations. By running synthetic personas through the blueprints, I gathered rapid qualitative feedback on layout friction and checkout trust before starting Figma production.
Conversational Trust & Retentive Flow
Designed a conversational WhatsApp flow focused on user retention and journey structure. Because the purchase experience was extremely fast, I carefully balanced micro-copy and confirmation touchpoints to deliver an agile yet highly secure feeling, ensuring users felt in control rather than rushed.
The Results
The Decolar AI Travel Assistant successfully compressed the booking funnel. By comparing service blueprints and running simulations with my own proprietary user-testing AI agent, I identified and resolved journey friction immediately, allowing the team to develop and ship a fully functional, high-retention MVP in just 30 days.
+90%
Flight Booking Velocity Increase
< 10min
Average Complete Checkout
Own AI-Agent
Synthetic Persona Simulator
30 Days
Fast MVP Launch
