Designing NxVoy: Redefining the Flight Booking UX with Conversational AI

Watch someone plan a trip sometime.Not book a flight — plan a trip.

They open Google Flights in one tab. Skyscanner in another. They check dates against their calendar. They open a weather site to see if it'll rain in Lisbon in March. They search "do I need a visa for Portugal" in a fourth tab. They check currency exchange rates. They look at airport transfer options. They read a Reddit thread about whether Terminal 2 at CDG has decent food for a three-hour layover.

By the time they actually book anything, they have eight tabs open, a dozen mental notes, and the lingering suspicion that they're overpaying.

This is the state of travel planning in 2024. Not broken — functional. But scattered, effortful, and exhausting for something that's supposed to be exciting.

NxVoy was designed around a simple provocation:

What if you could just say what you wanted — and get everything you needed in one place?

Not a search form. Not a filter panel. A conversation.

"Travel planning had become research. Eight tabs, five comparison sites, and three hours of your Saturday. NxVoy was designed to turn that research project back into a conversation."

What NxVoy wanted to be

NxVoy's long-term vision was ambitious: an all-in-one AI travel companion that handled flights, accommodations, attractions, group coordination, translation, travel support, and itinerary management.

Our engagement was more focused. We designed the foundation — the AI-powered flight planning and booking experience that would prove the concept and establish the platform's interaction model for everything that came after.

This was strategic. Flights are the hardest part of travel booking to get right. Complex routing, dynamic pricing, multi-airline combinations, cabin class variations, layover logistics — if the AI could handle flight planning conversationally while still supporting power users with traditional search, it could handle anything.

Get flights right, and the platform had permission to expand. Get flights wrong, and nothing else would matter.

Meet Sasha

The AI travel assistant wasn't a chatbot bolted onto a booking engine. Sasha was the primary interface — the first thing users encountered, and the experience the entire product was designed around.

A user could open NxVoy and type:

"Book a 3-day Dubai trip for 3 people on June 20 from Delhi."

And Sasha would:

1. Parse the intent — destination, dates, traveler count, origin

2. Search available flights across airlines and routes

3. Generate recommendations organized by priority

4. Provide a rich trip overview — weather, currency, visa requirements, local transport

5. Present bookable flight options with pricing and routing details

One sentence. One response. Everything the user needed to make a decision.

No forms. No filter panels. No "where from / where to / dates / passengers / cabin class" fields. A natural language request and a structured, actionable result.

What made Sasha different from a chatbot

The distinction matters.

A chatbot answers questions. "What's the cheapest flight to Dubai?" → "Here's a link to search results."

Sasha was a planning partner. It didn't just find flights — it understood travel context. A request for "a 3-day trip" implied specific outbound and return dates. "3 people" implied group booking considerations. "Dubai in June" triggered weather warnings (it's extremely hot) and contextual travel tips.

Sasha didn't wait to be asked. It anticipated what travelers needed to know — and presented it alongside the flights.

This was the fundamental interaction model: the AI doesn't just search. It plans. Every response was a miniature travel briefing, not just a list of results.

"Most travel AI answers questions. Sasha answered the question you asked and the five questions you didn't know to ask — visa requirements, weather conditions, airport transport, currency exchange — all in one response."

The hybrid model: AI-first, not AI-only

Here's the tension we had to resolve immediately:

Some travelers love conversational interfaces. They want to describe what they need in natural language, explore options through dialogue, and let the AI do the heavy lifting.

Other travelers — especially experienced ones — want manual control. They know their preferred airlines, they want specific filters, they have exact time preferences, and they find conversational interfaces slower than a well-designed search form.

Designing exclusively for either group would alienate the other.

NxVoy supported both:

Conversational planning

The AI-first experience. Users described their needs naturally, and Sasha translated intent into structured results.

This was the default entry point — the experience that differentiated NxVoy from every other travel platform. It was optimized for users who:

• Weren't sure exactly what they wanted

• Wanted guidance rather than just search results

• Were planning complex trips (multi-city, group, flexible dates)

• Valued convenience over control

Traditional flight search

For users who preferred manual control, a dedicated search flow supported:

• One-way trips — simple point-to-point booking

• Round trips — standard outbound and return

• Multi-city trips — complex itineraries with multiple legs

• Flexible traveler counts — solo to group

• Cabin class selection — economy through first class

• Advanced filtering — airlines, times, stops, duration, price range

The traditional search wasn't a fallback. It was a first-class experience — as polished and capable as any dedicated flight search platform.

The key design decision: both paths led to the same results. Whether a user found flights through Sasha or through manual search, the flight cards, comparison views, and booking flows were identical. The AI was an alternative entry point, not a separate product.

The trip overview: more than just flights

When Sasha generated flight recommendations, it didn't just show tickets. It provided a complete trip context — everything a traveler would normally research across multiple websites, consolidated into one view.

Flight details — routes, times, airlines, pricing, layover information.

Destination overview — what the place is like, what to expect, practical orientation.

Weather forecasts — current and projected conditions for the travel dates. No more googling "weather in Bali in August."

Visa guidance — do you need a visa? What's the process? Is it visa-on-arrival? This information changes the trip feasibility for many travelers and is usually discovered after booking.

Currency information — exchange rates, payment norms, tipping culture.

Local transportation — how to get from the airport to the city, public transit options, typical taxi costs.

Travel tips — practical, destination-specific advice that experienced travelers know and first-timers need.

This trip overview transformed the booking experience from "here are your flights" to "here's your trip." The flight was just one component of a richer travel briefing that helped users make more informed decisions.

"Traditional booking sites show flights. NxVoy showed trips. The flights were part of a richer picture that included everything you'd normally spend an hour researching separately."

Flight recommendations that reduce decisions

Decision fatigue is the silent killer of travel booking.

A search for "Delhi to Dubai, June 20" might return sixty results. Different airlines. Different times. Different layovers. Different prices. Scrolling through sixty options and evaluating each one against personal priorities — price vs. convenience vs. duration vs. airline preference — is cognitively expensive.

We organized flight recommendations around three clear priorities:

Best — the option that balanced price, duration, and convenience. The algorithmic recommendation for "if you don't want to think about it, pick this one."

Cheapest — the lowest fare option, regardless of convenience. For budget-conscious travelers who'll accept a 4 AM departure for a better price.

Fastest — the shortest total travel time. For business travelers and anyone who values their time over their wallet.

Three categories. Clear trade-offs. Instant orientation.

Users didn't have to evaluate sixty options. They evaluated three — each optimized for a different priority — and then explored variations within their preferred category.

This wasn't just sorting. It was curation. The platform made a judgment about what "best" meant and presented it confidently. Users could override that judgment with filters and manual exploration, but the starting point was always clear and opinionated.

The routing problem nobody sees

This section might seem technical. It's not. It's about trust.

Most travelers think booking a round-trip flight is simple: you fly out, you fly back. One airline, two flights, one ticket.

In reality, the airline routing ecosystem is extraordinarily complex. And NxVoy needed to support that complexity while making it understandable.

Standard round trips

The simple case. Same airline, outbound and return, single ticket. One set of baggage rules, one cancellation policy, one booking reference.

This is what most users expect. The design challenge was minimal — present the flights clearly and move toward checkout.

Hub routing

Multi-airline combinations connected through intermediary airports.

A Delhi-to-Dubai search might return a result that flies Delhi → Mumbai on one airline and Mumbai → Dubai on another. The connection happens at a hub airport, and the two flights might be operated by different carriers with different policies.

This required clear communication:

• Which airlines are involved at each leg?

• What's the connection time at the hub?

• Is baggage transferred automatically or does the traveler need to recheck?

• What happens if the first flight is delayed — is the connection protected?

The interface needed to show the routing logic — not just the flights. Users had to understand they were booking a connected journey, not two separate flights that happened to be sequential.

SOOWs (Sum of One-Ways)

The most complex — and most misunderstood — routing model.

A SOOW creates what looks like a round trip by combining two separate one-way tickets. The outbound and return are priced and ticketed independently.

Why does this matter? Because:

• Pricing might be lower — sometimes two one-ways cost less than a traditional round trip

• Baggage policies might differ — each one-way has its own baggage allowance

• Cancellation rules are separate — cancelling the outbound doesn't automatically cancel the return

• Airlines might be different — outbound on Airline A, return on Airline B

The design challenge: make this transparent without making it scary.

Most travelers don't know what a SOOW is. They don't need to. What they need is clear information: "This itinerary is priced as two separate tickets. Here's what that means for you." We designed inline explanations that surfaced when SOOWs appeared — brief, clear, and focused on practical implications (baggage, cancellation) rather than industry terminology.

"Airline routing is genuinely complex. Our job wasn't to simplify it — it was to translate it. Users didn't need to understand hub routing or SOOWs. They needed to understand what it meant for their trip."

The flight comparison experience

Once users identified promising options, they needed to evaluate them against each other.

The comparison view surfaced the dimensions that actually influence decisions:

• Total duration — including layovers, not just flight time

• Number of stops — direct vs. one-stop vs. multi-stop

• Airlines — carrier preferences and loyalty program considerations

• Layover details — where, how long, and whether it's a comfortable or rushed connection

• Total cost — fare plus any additional fees

• Cabin class — what's included at each price point

• Departure and arrival times — the 6 AM flight might be cheapest, but is it worth it?

The comparison view maintained trip context throughout — users never lost sight of the overall journey while evaluating individual flight options. This was a deliberate design decision: flight comparison shouldn't feel like a spreadsheet exercise detached from the trip it's part of.

Designing for mobile

Travel planning doesn't only happen at a desk.

People search flights on their commute. They compare prices while waiting in line. They check visa requirements from a café. They modify bookings from an airport lounge.

NxVoy needed to be fully capable on mobile — not a responsive adaptation that squeezed desktop layouts into a smaller screen, but a designed mobile experience that respected mobile interaction patterns.

What we designed for mobile

AI chat interactions — Sasha on mobile needed to feel like a messaging conversation, not a shrunken desktop interface. Chat bubbles, bottom-anchored input, contextual quick actions.

Flight search — the traditional search form redesigned for thumb-friendly input. Step-by-step instead of all-at-once. Origin, destination, dates, travelers — each on focused screens rather than cramped into one view.

Flight comparisons — cards instead of tables. Swipeable instead of scrollable. Key information first, details on expansion.

Trip summaries — the rich trip overview (weather, visa, currency) redesigned for vertical consumption. Sections that collapsed and expanded instead of dense horizontal layouts.

Booking flows — checkout optimized for mobile payment, with minimal typing and clear progress indicators.

Multiple mobile concepts were explored throughout the project — each iteration refining the balance between information density and mobile usability.

"We didn't make the desktop experience smaller. We redesigned travel planning for a screen you hold in one hand while standing on a train."

The trust problem with AI travel

This was the design challenge that ran underneath every feature in the platform:

How do you make users trust an AI with their travel plans?

Travel isn't a low-stakes domain. A wrong hotel recommendation wastes a few hundred dollars. A wrong flight recommendation wastes time, money, and potentially a trip. Missed visa requirements can mean being turned away at the border. Incorrect layover information can mean a missed connection.

The stakes are high. The tolerance for AI errors is low.

We addressed trust through several design strategies:

Show the reasoning, not just the result.

When Sasha recommended the "Best" flight, the interface showed why — balanced price, reasonable duration, well-reviewed airline, comfortable layover. Users could evaluate the recommendation criteria, not just the recommendation.

Source contextual information visibly.

Weather forecasts, visa requirements, and currency information included attribution and freshness indicators. "Visa information last updated 3 days ago" communicated that the data was maintained, not stale. Transparency about sources built confidence in accuracy.

Always offer manual verification.

Every AI-generated result could be verified through traditional search. If Sasha recommended a flight, users could open the same route in the manual search interface and confirm the options existed. The AI wasn't a black box — it was a faster path to the same results users could find themselves.

Be honest about limitations.

Sasha didn't pretend to know everything. If a query was outside its capabilities — complex multi-destination routing, niche airline partnerships, corporate booking requirements — it said so, and offered to connect users with the traditional search tools. Admitting limitations built more trust than feigning omniscience.

The design system: built for expansion

The design system for NxVoy was built with deliberate forward-thinking. Flights were the first module, but the platform was designed to eventually support hotels, attractions, activities, group planning, and full itinerary management.

This meant the design system couldn't be flight-specific. It had to be travel-generic — establishing patterns that would work for any travel product module.

Travel cards

The core display component. Used for flights in this engagement, but designed to work for hotels, attractions, activities, and any future travel entity. Consistent structure: primary information (what it is), key metrics (price, rating, duration), secondary details (expandable), and action (book, save, compare).

AI conversation patterns

Standardized patterns for how Sasha presented different types of information — flight results, trip overviews, destination information, follow-up questions. These patterns would extend to hotel recommendations, activity suggestions, and itinerary modifications in future modules.

Search modules

Form components, filter systems, and search result patterns designed to flex across different travel content types. The flight search form and the future hotel search form would share the same interaction patterns — date pickers, location inputs, traveler configurations, filter panels.

Flight result components

Specialized components for airline routing, layover display, cabin class presentation, and pricing breakdown. These were flight-specific but built on the generic card and comparison frameworks.

Comparison frameworks

Side-by-side evaluation patterns that worked for flights and would extend to hotels, restaurants, and activities. Consistent structure: entities as columns, attributes as rows, differences highlighted.

What was harder than expected

Translating natural language into bookable flights is deceptively complex.

"Book a 3-day Dubai trip" seems clear. But: does "3-day" mean 3 nights or 3 calendar days? Does "Dubai trip" mean flying into Dubai or any UAE airport? Does the user want to depart on June 20 or arrive on June 20? Natural language is ambiguous. Booking systems require precision.

We designed Sasha's responses to confirm interpretations before presenting results. "I found flights departing Delhi on June 20 and returning June 22 (2 nights). Is that right?" This micro-confirmation step prevented the most common frustration with AI travel tools: getting results for the wrong trip.

AI recommendations need to feel curated, not random.

When Sasha presented three priority categories (Best, Cheapest, Fastest), users expected those labels to mean something defensible. "Best" isn't just cheapest-plus-a-bit-more. It's a genuine evaluation of value — and explaining that evaluation without overwhelming users with algorithm details was a constant design calibration.

Routing complexity is invisible until it creates problems.

Most users never think about hub routing or SOOWs until something goes wrong — a missed connection, unexpected baggage fees, a cancellation policy they didn't understand. The challenge was surfacing routing information at the right moment: prominent enough that users noticed it, subtle enough that it didn't create unnecessary anxiety for simple bookings.

Mobile and desktop needed to be equally capable but differently designed.

Travel planning on mobile isn't a simplified version of desktop planning. It's a different context — different input methods, different attention spans, different environmental distractions. The mobile experience needed feature parity with desktop but interaction patterns designed for thumbs, interruptions, and variable connectivity.

The AI needed a personality without being annoying.

Sasha had to be helpful without being patronizing. Conversational without being chatty. Knowledgeable without being lecturing. Finding the right voice — warm but efficient, smart but not showy — required multiple iterations of conversation design that went far beyond visual UI.

Reflection

NxVoy was a project about changing the entry point of travel planning.

For twenty years, online travel booking has started the same way: a form. Where from, where to, when, how many. Fill it out, click search, scroll through results.

NxVoy started with a sentence. And that single shift — from form to conversation — changed everything downstream: how results were presented, how context was provided, how trust was established, and how decisions were made.

A few things this project taught us:

The best AI interfaces don't feel like AI interfaces.

Sasha didn't draw attention to being artificial intelligence. It felt like texting a knowledgeable friend who happened to have access to every airline's inventory. The technology was invisible. The helpfulness was visible. That's the right ratio.

Conversational UI is harder than traditional UI.

Forms are constrained. Users can only input what the form allows. Conversations are unconstrained — users can say anything, in any order, with any level of specificity. Designing for that openness — handling ambiguity, confirming interpretations, managing expectations — is more complex than designing the most sophisticated filter panel.

Travel is emotional, not transactional.

A flight search result is a transaction. A trip overview with weather, visa info, local tips, and destination context is the beginning of an experience. NxVoy's trip overview transformed the booking moment from "purchasing a ticket" to "imagining a journey." That emotional shift wasn't decorative — it was the product's primary differentiator.

Building the foundation matters more than building everything.

We designed flights. Not hotels. Not attractions. Not full itineraries. The scope was deliberately focused because getting the AI interaction model right for flights — the most complex travel product — established patterns that everything else could follow. The foundation was the product.

Trust in AI is earned through transparency and honesty.

Sasha earned trust by showing its reasoning, citing its sources, confirming its interpretations, and admitting its limitations. No amount of visual polish or conversation design could substitute for that fundamental transparency. Users trust AI that helps them make better decisions. They distrust AI that tries to make decisions for them.

What NxVoy became

NxVoy became the proof that travel planning could start with a sentence instead of a search form — and that the experience on the other side could be richer, more contextual, and more helpful than anything traditional travel platforms offered.

Not a chatbot added to a booking engine.

Not a search engine with an AI gimmick.

A genuinely new way to plan travel — where the AI understood intent, provided context, handled complexity, and let the traveler focus on what actually matters: where they're going and why.

The flights were the foundation. The conversation was the product. The trip was the experience.