Cruise Line Generative UI Case Study

Travel & Hospitality
Generative UI
Flutter GenUI SDK

See how VGV helped a global cruise line evolve a text-based AI assistant into a visual, generative concierge that moves guests closer to booking.

Booking conversion, driven by agentic search: 1.00% → 1.08%
Estimated ticketing revenue from that lift: $536M
Approved catalog items the model can assemble: 14

01At a glance

Charting a new course

01

The challenge

One of the world’s largest cruise lines had one of the first agentic assistants in cruise. But it answered in long paragraphs and deep-linked guests out of the experience to book.
02

Our approach

We guided the cruise line through Structured Creativity: user context, a catalog of approved UI components, and prompt rules that enforce behavior and brand voice. We then built a GenUI prototype so the team could experiment with the technology and test latency concerns.
03

The results

As part of the cruise line’s larger effort to build agentic discovery into its booking flow, booking conversion rose from 1.00% to 1.08%, an estimated $536 million in ticketing revenue. Our work gave the team a governed model for generative UI.

02The challenge

An early agentic leader, still answering in text

One of the world’s largest cruise lines moved early. Its AI assistant was one of the first agentic experiences in the cruise industry. Guests could ask rich planning questions and get real answers.

The answers were the problem. The assistant replied with long, information-dense paragraphs. When a guest was ready to act, it deep-linked into the existing website, pulling them out of the agentic experience and back into the same multi-step funnel every other guest used.

The cruise line wanted to explore how vacation discovery could become more visual. Could an LLM generate UI elements, not just text, that move prospective guests closer to booking?

In their words

“VGV is leading the charge in how a structured approach can help enterprises embrace GenUI.”

Group Product Manager, Google Seth Ladd

03Before and after

From a text-only agent to a visual, generative concierge

Text-based assistant answering a cruise booking request with long paragraphs of itineraries, balcony cabin prices and booking links — Before
Text-based agent

GenUI discovery flow from the prototype, filter chips for dates, destinations and cabin types above cards for cruises from California with ship photos, prices and View Details buttons — After VGV’s design and prototyping work
VGV envisions a GenUI-empowered discovery flow that uses visual cards, such as ship details and next best actions, to drive booking forward.

04Our approach

Listening before building

We started by listening. We met with the cruise line’s engineering, product, and design stakeholders to map constraints and find where GenUI could add value. We reviewed analytics on how the existing assistant flow affected booking conversion, and read real guest conversations to see what people actually ask.

We also worked directly with the Flutter team, which gave us early guidance on the Flutter GenUI SDK and a hand in shaping it.

05Structured Creativity

Three domains, one governed system

We guided the cruise line’s stakeholders through Structured Creativity, our process for designing GenUI that adapts to each guest without losing the brand. It works through three domains:

01
User context. What user context and data can we collect to shape each response?
02
Catalog. What is the visual catalog of UI components the model is allowed to use?
03
System prompt. What goes into the prompt to guide and enforce behavior and brand voice?

6 Message archetypes that shape every response

Structured Creativity

Structured Creativity balances helpful personalization with brand safety.

06Catalog

A catalog in the cruise line’s brand, fueled by context

The catalog is the set of pieces the model can assemble: destination cards, itinerary groups, ship comparisons, date pickers, step summaries, and more. Each item carries usage instructions, so the brand always looks and sounds like itself.

Context decides what to build for each guest. In the POC, it came from chat messages and a context bar that updates from the guest’s first request. Tapping a component writes the guest’s next message, so the conversation keeps moving without typing.

LLM Catalog slide for the button component from the Gen UI POC, its JSON schema in code beside the button variants it allows

07Architecture

An architecture built for speed

Latency was a real concern, so the technical architecture was a key part of the work. We designed GenUI as an agent layer on top of what the cruise line already runs. The existing assistant drafts its response, a GenUI model chooses the layout and catalog items, and the Flutter GenUI SDK renders interactive UI in real time.

Because the layer sits beside the assistant’s backend rather than inside it, any of the cruise line’s product teams can plug in the same capability. One Dart and Flutter codebase renders across web, mobile, and kiosk.

Google · Flutter & GenUI collaboration

Built alongside the team behind Flutter and GenUI

VGV works directly with Google’s Flutter team. On this engagement, that collaboration gave us early guidance on the Flutter GenUI SDK and a hand in shaping it. Google also worked with the cruise line and VGV to uncover and test latency in the prototype.

08The prototype

Prototyping to explore latency issues

VGV delivered a working prototype that used GenUI and low-fidelity UI elements as a step toward the cruise line’s ultimate vision.

The prototype also helped the cruise line, VGV, and Google uncover latency issues. Some visual responses took up to 30 seconds to arrive from the LLM. As of late 2026, VGV has brought those latency benchmarks under 4 seconds.

09Where this goes next

Beyond booking

Booking was the starting point, not the boundary. VGV helped the cruise line conceptualize how the same governed system of system prompt, catalog, and context could shape the rest of the cruise experience, from choosing a sailing to ordering onboard to exploring a port from the cabin.

Each concept runs on the same three inputs as the POC. That gives the cruise line one model for adding AI to guest experiences that stays on-brand, governable, and portable across every screen it owns.

Search that knows you

Results adapt to past cruises, interests, and the whole travel party. Each result explains why it appears, and guests see a short set of top matches instead of a long list.
A kiosk that recognizes regulars

With opt-in recognition, an onboard kiosk filters the menu for dietary needs and surfaces favorites. The layout rebuilds around the guest’s history, and one tap completes the order.
A cabin window that becomes a guide

As the ship nears the coast, the window becomes an interactive display. It layers hotspots, wildlife, and excursions over the real view, and a tap starts a booking in the concierge chat.

Roll credits.

Who did what, and with which tools.

Client: One of the world’s largest cruise lines
Services engaged: Product StrategyExperience DesignGenerative UIDart & Flutter Development
Worked with: The Flutter teamGoogle
Built with: Flutter GenUI SDKDart and Flutter
Delivered: Structured CreativityA working GenUI prototype