“ Spur ”
Transforming travel planning into a collaborative dialogue, from conversation to itinerary in under 3 minutes
Date
Nov 2025
Tools
Figma, Midjourney,
Claude AI, Notion AI
Focus
AI-native product design, conversational
interfaces, multi-modal interaction
*
THE PROBLEM
Traditional travel planning is exhausting.
Current travel booking experiences force users
into rigid workflows:
Filter fatigue: Users must define preferences across 20+ filters before seeing relevant results
Decision paralysis: Overwhelming choice with minimal guidance leads to abandoned bookings
Fragmented research: Planning requires juggling multiple tabs, apps, and spreadsheets
Time-intensive: Average trip planning takes 5+ hours of active research
Traditional platforms demand structured input before discovery.
(e.g. Expedia)
💡
The opportunity:
What if planning a trip felt as natural as describing it to a friend?
Advances in conversational AI enable interfaces that understand context, ask clarifying questions, and learn from preference patterns, creating experiences that feel collaborative rather than transactional.
*
COMPETITIVE ANALYSIS
I analyzed how current platforms (Expedia, Airbnb Experiences, Kayak, TripIt) handle trip planning …
Key findings:
Most platforms optimize for booking rather than discovering, they assume users already know where/when they want to go
Filter-based search requires users to already understand their preferences in platform-specific categories
Zero learning between trips, users must re-input preferences every time
Itinerary tools are "dumb containers" that don't offer intelligent sequencing or optimization
The gap:
No platform treats discovery and booking as a unified, conversational flow.
*
USER RESEARCH
I conducted interviews with 8 frequent travelers (4+ trips/year) to understand their planning process.
Three distinct personas emerged:
The Impulsive Explorer
Makes trip decisions quickly based on "vibe" rather than exhaustive research
Frustrated by platforms requiring dates/budget before seeing inspiration
Wants: "Just show me something cool I can afford"
"I know I want to go somewhere, but I don't know where. Every app wants me to pick a destination first, that's what I'm trying to figure out!"
— Sarah, 28, Marketing Manager
The Methodical Planner
Spends hours researching, comparing options, building detailed spreadsheets
Trusts their own research over algorithms
Wants: Tools that organize their research, not replace it
"I'm not against technology helping me, but I need to see why it's suggesting things. Show me your work."
— James, 35, Software Engineer
The Delegator
Would rather someone else handle logistics entirely
Uses travel agents or asks well-traveled friends for recommendations
Wants: Expertise without the premium price tag
"I just want someone to tell me: 'Go here, do this, it'll be great.' I don't have time to become a travel expert."
— Linda, 42, Executive
*
DESIGNING FOR AI
Before diving into interface design, I established principles for creating AI-native experiences that feel genuinely helpful rather than gimmicky:
Conversation over forms
AI should reduce friction, not add steps. Every form field replaced with natural language is cognitive load removed.
Progressive disclosure through dialogue
Rather than overwhelming users with options upfront, let AI ask questions contextually as understanding deepens.
Transparency in AI reasoning
Users should understand why the AI made specific recommendations. Show the match logic, not just the match score.
Graceful degradation
When AI confidence is low, surface this uncertainty honestly and offer alternatives rather than confidently presenting mediocre results.
Human override at every turn
AI suggestions should always be starting points, not mandates. Make it trivially easy to modify, regenerate, or ignore AI output.
*
DESIGN CONSTRAINTS
& TRADEOFFS
Technical Constraints:
Voice accuracy
Speech recognition degrades in noisy environments → designed text fallback as equal-priority input method
Price accuracy
Real-time pricing from all providers isn't feasible → matched estimates with booking platform APIs where available, clearly labeled "estimated" elsewhere
Data availability
Not all destinations have comprehensive activity data → AI trained to acknowledge gaps rather than hallucinate options
Deliberate Tradeoffs:
Transparency depth vs. cognitive load
The 92% match badge shows 4 key factors. Testing revealed 4 was the sweet spot (3 felt incomplete, 5+ felt overwhelming)
AI confidence display
Chose to show match percentage for all results rather than hiding low-confidence options. Users preferred seeing 60% matches with explanations over mysteriously limited results
Onboarding vs. immediacy
Skipped traditional onboarding tutorial in favor of example prompts. Lost some feature discoverability but gained 40% faster time-to-first-action
“ Spur ”
Transforming travel planning into a collaborative dialogue, from conversation to itinerary in under 3 minutes
Date
Nov 2025
Tools
Figma, Midjourney,
Claude AI, Notion AI
Focus
AI-native product design, conversational
interfaces, multi-modal interaction
*
THE PROBLEM
Traditional travel planning is exhausting.
Current travel booking experiences force users
into rigid workflows:
Filter fatigue: Users must define preferences across 20+ filters before seeing relevant results
Decision paralysis: Overwhelming choice with minimal guidance leads to abandoned bookings
Fragmented research: Planning requires juggling multiple tabs, apps, and spreadsheets
Time-intensive: Average trip planning takes 5+ hours of active research
Traditional platforms demand structured input before discovery.
(e.g. Expedia)
💡
The opportunity:
What if planning a trip felt as natural as describing it to a friend?
Advances in conversational AI enable interfaces that understand context, ask clarifying questions, and learn from preference patterns, creating experiences that feel collaborative rather than transactional.
*
COMPETITIVE ANALYSIS
I analyzed how current platforms (Expedia, Airbnb Experiences, Kayak, TripIt) handle trip planning …
Key findings:
Most platforms optimize for booking rather than discovering, they assume users already know where/when they want to go
Filter-based search requires users to already understand their preferences in platform-specific categories
Zero learning between trips, users must re-input preferences every time
Itinerary tools are "dumb containers" that don't offer intelligent sequencing or optimization
The gap:
No platform treats discovery and booking as a unified, conversational flow.
*
USER RESEARCH
I conducted interviews with 8 frequent travelers (4+ trips/year) to understand their planning process.
Three distinct personas emerged:
The Impulsive Explorer
Makes trip decisions quickly based on "vibe" rather than exhaustive research
Frustrated by platforms requiring dates/budget before seeing inspiration
Wants: "Just show me something cool I can afford"
"I know I want to go somewhere, but I don't know where. Every app wants me to pick a destination first, that's what I'm trying to figure out!"
— Sarah, 28, Marketing Manager
The Methodical Planner
Spends hours researching, comparing options, building detailed spreadsheets
Trusts their own research over algorithms
Wants: Tools that organize their research, not replace it
"I'm not against technology helping me, but I need to see why it's suggesting things. Show me your work."
— James, 35, Software Engineer
The Delegator
Would rather someone else handle logistics entirely
Uses travel agents or asks well-traveled friends for recommendations
Wants: Expertise without the premium price tag
"I just want someone to tell me: 'Go here, do this, it'll be great.' I don't have time to become a travel expert."
— Linda, 42, Executive
*
DESIGNING FOR AI
Before diving into interface design, I established principles for creating AI-native experiences that feel genuinely helpful rather than gimmicky:
Conversation over forms
AI should reduce friction, not add steps. Every form field replaced with natural language is cognitive load removed.
Progressive disclosure through dialogue
Rather than overwhelming users with options upfront, let AI ask questions contextually as understanding deepens.
Transparency in AI reasoning
Users should understand why the AI made specific recommendations. Show the match logic, not just the match score.
Graceful degradation
When AI confidence is low, surface this uncertainty honestly and offer alternatives rather than confidently presenting mediocre results.
Human override at every turn
AI suggestions should always be starting points, not mandates. Make it trivially easy to modify, regenerate, or ignore AI output.
*
DESIGN CONSTRAINTS
& TRADEOFFS
Technical Constraints:
Voice accuracy
Speech recognition degrades in noisy environments → designed text fallback as equal-priority input method
Price accuracy
Real-time pricing from all providers isn't feasible → matched estimates with booking platform APIs where available, clearly labeled "estimated" elsewhere
Data availability
Not all destinations have comprehensive activity data → AI trained to acknowledge gaps rather than hallucinate options
Deliberate Tradeoffs:
Transparency depth vs. cognitive load
The 92% match badge shows 4 key factors. Testing revealed 4 was the sweet spot (3 felt incomplete, 5+ felt overwhelming)
AI confidence display
Chose to show match percentage for all results rather than hiding low-confidence options. Users preferred seeing 60% matches with explanations over mysteriously limited results
Onboarding vs. immediacy
Skipped traditional onboarding tutorial in favor of example prompts. Lost some feature discoverability but gained 40% faster time-to-first-action