Designing a conversational AI experience for smarter travel
Role
Product Designer
Timeline
1 week
Client
Airbnb
Industry
Travel
Responsabilities
UX audit for mobile app
Research and discovery process
Information architecture
Competitor analysis
Problem analysis and diagnosis
UX strategy for the Brazilian market
Overview
A concept project exploring how conversational AI can enhance the Airbnb experience by simplifying trip planning, reducing stress, and making the product more inclusive. The goal was to design a feature that helps users plan and book trips seamlessly while reflecting Airbnb’s friendly, human-centered brand tone.
Challenges
Unclear hierarchy of actions
High cognitive load for users
Fragmented navigation across modules
Low task efficiency
Difficulty scaling new features
Inconsistent terminology and patter
Poor scalability for new features
Context and challenge
This project started from a feature request, not from a problem statement. The challenge was to envision how a conversational interface could be integrated into Airbnb’s ecosystem to make travel planning simpler, more engaging, and accessible to a broader audience.
Market data revealed that nearly 49% of travelers feel stressed during trip planning, and 65% of older adults face difficulties using complex digital tools. Meanwhile, competitors such as Booking.com and Hopper were launching AI-powered assistants to guide users through search and booking flows.
Primary goals
Help users plan trips with less stress
Simplify complex decision-making through natural conversation
Expand Airbnb’s accessibility to older audiences
Enhance user satisfaction with a friendlier, more human-centered flow
Stay consistent with Airbnb’s architecture and visual identity
Research and data analysis
To understand the opportunity, I analyzed Airbnb’s current user journeys, design system, and information architecture, identifying key friction points in the search and booking flows. Users often felt overwhelmed by too many options, unclear filters, and multiple decision points that demanded high cognitive effort.
From the user perspective, travel planning was stressful and complex. From the stakeholder perspective, Airbnb needed innovation to remain competitive in a market rapidly embracing AI. And from the market perspective, data showed a growing openness to AI tools for travel planning and personalization.
These insights revealed a clear opportunity: build an AI-powered assistant that could make the process conversational, human, and less overwhelming.
Diagnosis of the scenario
Before defining the concept, I analyzed Airbnb’s ecosystem to identify where a conversational assistant could create real value. Users often felt overwhelmed by too many options, filters, and decisions, leading to cognitive overload and early drop-offs during trip planning.
Competitors like Booking and Hopper were already introducing AI tools, but their focus was mostly on automation rather than emotion. The opportunity for Airbnb was to design a human-centered AI experience that simplified travel planning while preserving the brand’s warm, personal tone.
This diagnosis revealed that the solution needed to balance two dimensions: the emotional side of travel, curiosity and anticipation, and the functional side, speed, clarity, and efficiency.
Main problems
Overwhelming number of filters and property options
Users struggling with indecision and information overload
Travel planning perceived as stressful and time-consuming
Limited accessibility for elderly and users with disabilities
Lack of personalized guidance during trip discovery
Benchmark and insights
The benchmark included Booking.com, Skyscanner, and Hopper, which already integrated AI to predict prices and provide recommendations.
Key insights emerged:
Conversational AI can simplify multi-step tasks.
Natural language fosters engagement and emotional connection.
Dynamic visual components help reduce monotony and guide focus.
Personalization builds trust and increases conversion.
These findings helped define the conceptual direction for Airbnb’s own assistant.
Designing an AI experience
Designing an AI experience required a more flexible approach to journey mapping. Instead of a single linear flow, I built adaptive conversation trees that could respond to changing user intent and different entry points.
Mapping this experience also required anticipating the unpredictable: how users might skip steps, change destinations mid-conversation, or ask open-ended questions. The structure was designed to adapt gracefully, maintaining flow even when user intent shifted.
The assistant’s personality (“Ava”) was defined to reflect Airbnb’s values, friendly, inclusive, and empathetic, using natural tone and conversational pacing to make users feel guided rather than interrogated.
Accessibility was also key. Text readability, contrast, and clarity of instructions were refined to ensure older adults and less tech-savvy users could engage easily. The result was a conversational experience that felt natural, adaptive, and truly human.
Pillars of the sollution
Natural Interaction
A conversational interface that understands user intent, adapts to context, and uses human-like language
Hyper Tailoring
AI that tailors suggestions and recommendations based on minimal user input and past behavior
Dynamic Components
Visual cards integrated into the chat to summarize key data (destinations, prices,..) without overwhelming text
Friendly Chat
Introducing “Ava,” a virtual travel assistant with a warm, professional, and approachable tone aligned with Airbnb’s brand.
Design solution
Natural Interaction:
Making AI feel human
The foundation of the design was to create a conversational interface that feels natural and intuitive. The assistant interprets user intent, adapts to context, and communicates with human-like language to reduce friction.
Instead of users navigating filters and lists, the conversation guides them step by step, asking about budget, destination, and preferences in a friendly tone. This not only simplifies the process but also helps users focus on one task at a time, reducing cognitive effort and making the experience approachable for older audiences.
Friendly Chat: Designing Ava’s personality
At the heart of the experience is Ava, Airbnb’s virtual travel assistant, a friendly and approachable persona that embodies the brand’s voice.
She communicates with empathy and clarity, guiding users through choices and reducing stress during trip planning.
When a question is unclear, Ava offers gentle guidance instead of error messages, keeping the experience human, simple, and fluid.
Hyper Tailoring: Personalization that feels effortless
AI-driven tailoring ensures every interaction is adapted to each user’s reality. Based on minimal input, the assistant adjusts tone, pacing, and recommendations dynamically, offering personalized results for solo travelers, families, or digital nomads.
By learning from behavior and preferences, it refines suggestions over time, helping users quickly find stays and experiences that truly match their expectations. This made the journey more inclusive, responsive, and efficient.
Dynamic Components: Bringing clarity to the flow
To break the monotony of text-based interactions, dynamic visual cards were introduced. These cards summarize key data, such as destination, price, and amenities, and allow users to make quick, informed choices without scrolling or switching contexts.
Predictive responses integrated into these components also help users act faster, transforming yes/no answers or quick refinements into seamless actions. This balance of visual and conversational design made the flow more engaging and reduced decision fatigue.
Usability testing
%
new convertion rate
%
users asked for improvement in labels and feedback
%
main tasks of the exploration journey was completed
The results showed me areas where I should improve text accessibility and create clearer feedback. In the end, the screens were refined.
Next steps
MVP negotiation
A high-fidelity prototype was built in Figma and tested with five users aged 30–65, mostly occasional travelers. The usability tests revealed improvements needed in text accessibility, visual feedback, and clarity of conversational responses. After refinements, users described the flow as “clear,” “friendly,” and “less stressful.”
Together with the PM and tech lead, we defined an MVP scope focused on:
The conversational flow with “Ava.”
Trip discovery and booking assistance.
The option to save and share itineraries.
This MVP was designed to validate whether conversational AI could reduce stress and increase booking conversion before scaling to other areas of the app.
Three product stages
The next steps would include tracking KPIs such as booking conversion rates, exploration triggers, and time spent on planning versus traditional flows.
Based on usability data, we planned an iterative roadmap:
Short term: Refine accessibility and conversational feedback.
Mid term: Integrate predictive elements such as price alerts and route visualization.
Long term: Expand “Ava” across the platform and introduce multi-user collaboration for group trip planning.
In retrospect
This project reinforced how conversational interfaces can transform complex journeys into intuitive, human-centered experiences. It showed that AI-driven design is most valuable when it enhances clarity, not complexity, and when it adapts naturally to how users think and speak.
Beyond the prototype, the greatest learning was understanding how UX, AI, and architecture must work together, to make the product not only smart but also empathetic, inclusive, and aligned with the brand’s essence of connection and belonging.
Designing a conversational AI experience for smarter travel
Timeline
1 week
Client
Airbnb
Industry
Travel
Role
Product Designer
Responsibilities
UX audit for mobile app
Research and discovery process
Information architecture
Competitor analysis
Problem analysis and diagnosis
UX strategy for the Brazilian market
Overview
A concept project exploring how conversational AI can enhance the Airbnb experience by simplifying trip planning, reducing stress, and making the product more inclusive. The goal was to design a feature that helps users plan and book trips seamlessly while reflecting Airbnb’s friendly, human-centered brand tone.
Challenges
Unclear hierarchy of actions
High cognitive load for users
Fragmented navigation across modules
Low task efficiency
Difficulty scaling new features
Inconsistent terminology and patter
Poor scalability for new features
Context and challenge
This project started from a feature request, not from a problem statement. The challenge was to envision how a conversational interface could be integrated into Airbnb’s ecosystem to make travel planning simpler, more engaging, and accessible to a broader audience.
Market data revealed that nearly 49% of travelers feel stressed during trip planning, and 65% of older adults face difficulties using complex digital tools. Meanwhile, competitors such as Booking.com and Hopper were launching AI-powered assistants to guide users through search and booking flows.
Primary goals
Help users plan trips with less stress
Simplify complex decision-making through natural conversation
Expand Airbnb’s accessibility to older audiences
Enhance user satisfaction with a friendlier, more human-centered flow
Stay consistent with Airbnb’s architecture and visual identity
Research and data analysis
To understand the opportunity, I analyzed Airbnb’s current user journeys, design system, and information architecture, identifying key friction points in the search and booking flows. Users often felt overwhelmed by too many options, unclear filters, and multiple decision points that demanded high cognitive effort.
From the user perspective, travel planning was stressful and complex. From the stakeholder perspective, Airbnb needed innovation to remain competitive in a market rapidly embracing AI. And from the market perspective, data showed a growing openness to AI tools for travel planning and personalization.
These insights revealed a clear opportunity: build an AI-powered assistant that could make the process conversational, human, and less overwhelming.
Diagnosis of the scenario
Before defining the concept, I analyzed Airbnb’s ecosystem to identify where a conversational assistant could create real value. Users often felt overwhelmed by too many options, filters, and decisions, leading to cognitive overload and early drop-offs during trip planning.
Competitors like Booking and Hopper were already introducing AI tools, but their focus was mostly on automation rather than emotion. The opportunity for Airbnb was to design a human-centered AI experience that simplified travel planning while preserving the brand’s warm, personal tone.
This diagnosis revealed that the solution needed to balance two dimensions: the emotional side of travel, curiosity and anticipation, and the functional side, speed, clarity, and efficiency.
Main problems
Overwhelming number of filters and property options
Users struggling with indecision and information overload
Travel planning perceived as stressful and time-consuming
Limited accessibility for elderly and users with disabilities
Lack of personalized guidance during trip discovery
Pillars of the sollution
Natural Interaction
A conversational interface that understands user intent, adapts to context, and uses human-like language
Hyper Tailoring
AI that tailors suggestions and recommendations based on minimal user input and past behavior
Dynamic Components
Visual cards integrated into the chat to summarize key data (destinations, prices,..) without overwhelming text
Friendly Chat
Introducing “Ava,” a virtual travel assistant with a warm, professional, and approachable tone aligned with Airbnb’s brand.
Benchmark and insights
The benchmark included Booking.com, Skyscanner, and Hopper, which already integrated AI to predict prices and provide recommendations.
Key insights emerged:
Conversational AI can simplify multi-step tasks.
Natural language fosters engagement and emotional connection.
Dynamic visual components help reduce monotony and guide focus.
Personalization builds trust and increases conversion.
These findings helped define the conceptual direction for Airbnb’s own assistant.
Designing an AI experience
Designing an AI experience required a more flexible approach to journey mapping. Instead of a single linear flow, I built adaptive conversation trees that could respond to changing user intent and different entry points.
Mapping this experience also required anticipating the unpredictable: how users might skip steps, change destinations mid-conversation, or ask open-ended questions. The structure was designed to adapt gracefully, maintaining flow even when user intent shifted.
The assistant’s personality (“Ava”) was defined to reflect Airbnb’s values, friendly, inclusive, and empathetic, using natural tone and conversational pacing to make users feel guided rather than interrogated.
Accessibility was also key. Text readability, contrast, and clarity of instructions were refined to ensure older adults and less tech-savvy users could engage easily. The result was a conversational experience that felt natural, adaptive, and truly human.
Design solution
Natural Interaction: Making AI feel human
The foundation of the design was to create a conversational interface that feels natural and intuitive. The assistant interprets user intent, adapts to context, and communicates with human-like language to reduce friction.
Instead of users navigating filters and lists, the conversation guides them step by step, asking about budget, destination, and preferences in a friendly tone. This not only simplifies the process but also helps users focus on one task at a time, reducing cognitive effort and making the experience approachable for older audiences.
Friendly Chat: Designing Ava’s personality
At the heart of the experience is Ava, Airbnb’s virtual travel assistant, a friendly and approachable persona that embodies the brand’s voice.
She communicates with empathy and clarity, guiding users through choices and reducing stress during trip planning.
When a question is unclear, Ava offers gentle guidance instead of error messages, keeping the experience human, simple, and fluid.
Hyper Tailoring: Personalization that feels effortless
AI-driven tailoring ensures every interaction is adapted to each user’s reality. Based on minimal input, the assistant adjusts tone, pacing, and recommendations dynamically, offering personalized results for solo travelers, families, or digital nomads.
By learning from behavior and preferences, it refines suggestions over time, helping users quickly find stays and experiences that truly match their expectations. This made the journey more inclusive, responsive, and efficient.
Dynamic Components: Bringing clarity to the flow
To break the monotony of text-based interactions, dynamic visual cards were introduced. These cards summarize key data, such as destination, price, and amenities, and allow users to make quick, informed choices without scrolling or switching contexts.
Predictive responses integrated into these components also help users act faster, transforming yes/no answers or quick refinements into seamless actions. This balance of visual and conversational design made the flow more engaging and reduced decision fatigue.
Usability testing
%
new convertion rate
%
users asked for improvement in labels and feedback
%
main tasks of the exploration journey was completed
The results showed me areas where I should improve text accessibility and create clearer feedback. In the end, the screens were refined.
Next steps
MVP negotiation
During usability testing, a few key improvement opportunities were identified. Some users still struggled to locate secondary actions within nested menus, while others suggested clearer visual feedback when switching between modules. There was also a need to improve the discoverability of shortcuts and to refine certain labels that caused hesitation during navigation.
After analyzing these findings, the design, product, and tech leads worked together to define an MVP scope that prioritized solving the most critical issues without delaying release. The MVP included the new navigation structure, improved hierarchy, and contextual shortcuts, focusing on the most frequent user flows such as bill payments, document access, and resident communication.
Three product stages
The next steps would include tracking KPIs such as booking conversion rates, exploration triggers, and time spent on planning versus traditional flows.
Based on usability data, we planned an iterative roadmap:
Short term: Refine accessibility and conversational feedback.
Mid term: Integrate predictive elements such as price alerts and route visualization.
Long term: Expand “Ava” across the platform and introduce multi-user collaboration for group trip planning.
In retrospect
This project reinforced how conversational interfaces can transform complex journeys into intuitive, human-centered experiences. It showed that AI-driven design is most valuable when it enhances clarity, not complexity, and when it adapts naturally to how users think and speak.
Beyond the prototype, the greatest learning was understanding how UX, AI, and architecture must work together, to make the product not only smart but also empathetic, inclusive, and aligned with the brand’s essence of connection and belonging.
Designing a conversational AI experience for smarter travel
Role
Product Designer
Timeline
1 week
Client
Airbnb
Industry
Travel
Responsibilities
UX audit for mobile app
Research and discovery process
Information architecture
Competitor analysis
Problem analysis and diagnosis
UX strategy for the Brazilian market
Overview
A concept project exploring how conversational AI can enhance the Airbnb experience by simplifying trip planning, reducing stress, and making the product more inclusive. The goal was to design a feature that helps users plan and book trips seamlessly while reflecting Airbnb’s friendly, human-centered brand tone.
Challenges
Unclear hierarchy of actions
High cognitive load for users
Fragmented navigation across modules
Low task efficiency
Difficulty scaling new features
Inconsistent terminology and patter
Poor scalability for new features
Context and challenge
This project started from a feature request, not from a problem statement. The challenge was to envision how a conversational interface could be integrated into Airbnb’s ecosystem to make travel planning simpler, more engaging, and accessible to a broader audience.
Market data revealed that nearly 49% of travelers feel stressed during trip planning, and 65% of older adults face difficulties using complex digital tools. Meanwhile, competitors such as Booking.com and Hopper were launching AI-powered assistants to guide users through search and booking flows.
Primary goals
Help users plan trips with less stress
Simplify complex decision-making through natural conversation
Expand Airbnb’s accessibility to older audiences
Enhance user satisfaction with a friendlier, more human-centered flow
Stay consistent with Airbnb’s architecture and visual identity
Research and data analysis
To understand the opportunity, I analyzed Airbnb’s current user journeys, design system, and information architecture, identifying key friction points in the search and booking flows. Users often felt overwhelmed by too many options, unclear filters, and multiple decision points that demanded high cognitive effort.
From the user perspective, travel planning was stressful and complex. From the stakeholder perspective, Airbnb needed innovation to remain competitive in a market rapidly embracing AI. And from the market perspective, data showed a growing openness to AI tools for travel planning and personalization.
These insights revealed a clear opportunity: build an AI-powered assistant that could make the process conversational, human, and less overwhelming.
Diagnosis of the scenario
Before defining the concept, I analyzed Airbnb’s ecosystem to identify where a conversational assistant could create real value. Users often felt overwhelmed by too many options, filters, and decisions, leading to cognitive overload and early drop-offs during trip planning.
Competitors like Booking and Hopper were already introducing AI tools, but their focus was mostly on automation rather than emotion. The opportunity for Airbnb was to design a human-centered AI experience that simplified travel planning while preserving the brand’s warm, personal tone.
This diagnosis revealed that the solution needed to balance two dimensions: the emotional side of travel, curiosity and anticipation, and the functional side, speed, clarity, and efficiency.
Main problems
Overwhelming number of filters and property options
Users struggling with indecision and information overload
Travel planning perceived as stressful and time-consuming
Limited accessibility for elderly and users with disabilities
Lack of personalized guidance during trip discovery
Benchmark and insights
The benchmark included Booking.com, Skyscanner, and Hopper, which already integrated AI to predict prices and provide recommendations.
Key insights emerged:
Conversational AI can simplify multi-step tasks.
Natural language fosters engagement and emotional connection.
Dynamic visual components help reduce monotony and guide focus.
Personalization builds trust and increases conversion.
These findings helped define the conceptual direction for Airbnb’s own assistant.
Pillars of the sollution
Natual Interaction
A conversational interface that understands user intent, adapts to context, and uses human-like language
Hyper Tailoring
AI that tailors suggestions and recommendations based on minimal user input and past behavior
Dynamic Components
Visual cards integrated into the chat to summarize key data (destinations, prices,..) without overwhelming text
Friendly Chat
Introducing “Ava,” a virtual travel assistant with a warm, professional, and approachable tone aligned with Airbnb’s brand.
Design solution
Simplicity: Making navigation effortless
In addition to reducing the number of screens, this page was designed to centralize all user information required during onboarding.
Once the SMS verification is completed, Gruvi checks if there is a property manager in the system with the data provided by the user.
If no matching record is found, the user immediately sees the message “No property found” at step 3 of the flow, saving time and avoiding unnecessary steps.
Relevance: Showing what truly matters
Another important change was the introduction of pending action banners, allowing tasks like e-mail confirmation or identity validation to be postponed.
This way, users could explore the app and access their condominium information immediately, while still being reminded of the steps they needed to complete.
Tailoring: Designing for each user’s reality
Property information is displayed clearly and in a visually friendly way, acting as a positive confirmation during onboarding.
This avoids redundant steps and builds confidence by immediately showing users that they are accessing the correct community within Gruvi.
This type of delight is not merely aesthetic but functional, t reduces uncertainty, reinforces trust, and creates a welcoming moment that prepares users to continue their journey with confidence.
Prediction: Anticipating needs before they arise
The “no match found” screen displays a message informing the user that no property is linked to the provided information and that they should request data correction from their property management company.
We also offered the option to try again with different information in case of a typing error, as well as the possibility to request an update of their registration directly with the condominium management.
Usability testing
%
users asked for improvement in labels and feedback
%
tasks of the exploration journey was completed
The results showed me areas where I should improve text accessibility and create clearer feedback. In the end, the screens were refined.
Next steps
Designing an AI experience
Designing an AI experience required a more flexible approach to journey mapping. Instead of a single linear flow, I built adaptive conversation trees that could respond to changing user intent and different entry points.
Mapping this experience also required anticipating the unpredictable: how users might skip steps, change destinations mid-conversation, or ask open-ended questions. The structure was designed to adapt gracefully, maintaining flow even when user intent shifted.
The assistant’s personality (“Ava”) was defined to reflect Airbnb’s values, friendly, inclusive, and empathetic, using natural tone and conversational pacing to make users feel guided rather than interrogated.
Accessibility was also key. Text readability, contrast, and clarity of instructions were refined to ensure older adults and less tech-savvy users could engage easily. The result was a conversational experience that felt natural, adaptive, and truly human.
MVP negotiation
During usability testing, a few key improvement opportunities were identified. Some users still struggled to locate secondary actions within nested menus, while others suggested clearer visual feedback when switching between modules. There was also a need to improve the discoverability of shortcuts and to refine certain labels that caused hesitation during navigation.
After analyzing these findings, the design, product, and tech leads worked together to define an MVP scope that prioritized solving the most critical issues without delaying release. The MVP included the new navigation structure, improved hierarchy, and contextual shortcuts, focusing on the most frequent user flows such as bill payments, document access, and resident communication.
Three product stages
The next steps would include tracking KPIs such as booking conversion rates, exploration triggers, and time spent on planning versus traditional flows.
Based on usability data, we planned an iterative roadmap:
Short term: Refine accessibility and conversational feedback.
Mid term: Integrate predictive elements such as price alerts and route visualization.
Long term: Expand “Ava” across the platform and introduce multi-user collaboration for group trip planning.
In retrospect
This project reinforced how conversational interfaces can transform complex journeys into intuitive, human-centered experiences. It showed that AI-driven design is most valuable when it enhances clarity, not complexity, and when it adapts naturally to how users think and speak.
Beyond the prototype, the greatest learning was understanding how UX, AI, and architecture must work together, to make the product not only smart but also empathetic, inclusive, and aligned with the brand’s essence of connection and belonging.