Kitravia
Crafting an AI-Powered "Agentic Travel" Ecosystem
Overview
Kitravia was an ambitious attempt to redesign travel booking around an AI-assisted, conversational experience rather than the usual form-heavy flow. The product vision was to help users move from fragmented searching, comparing, and validating across multiple tabs toward a single guided journey that felt faster, clearer, and more trustworthy.
This project focused on turning that vision into a credible end-to-end B2C experience. The work covered conversational search, search results, a structured five-step booking flow, and a dedicated visa booking hub, with the broader aim of reducing booking friction while keeping users in control.
Product Context
Most travel platforms still rely on rigid forms, repetitive filtering, and manual comparison behavior. Research showed that users often default to familiar platforms such as Booking.com not because those products are exciting, but because they are predictable, usable, and trusted.
That context created both the opportunity and the challenge for Kitravia. To stand out, the product could not simply imitate the category leader; it had to offer a meaningfully better planning and booking experience while overcoming the trust barrier that naturally appears when AI enters a high-stakes flow like travel booking and payment.
Role and Scope
The redesign was completed in a team of three. Within that team, the design lead owned the primary B2C flow, including conversational AI search, search results, the five-step booking flow, and the visa booking hub.
The scope went beyond visual cleanup. It required reshaping information architecture, interaction patterns, trust moments, and mobile-first entry points so the product felt understandable on first contact and scalable enough for handoff to development.
Goals
The redesign aimed to replace a manual, multi-tab booking process with an intelligent concierge-style experience. A key project objective was a projected 40 percent reduction in average booking time through a more guided and conversational flow.
From a UX perspective, the goals were more specific:
Reduce cognitive load during search and decision-making.
Make the AI interaction feel understandable rather than opaque.
Preserve user confidence through explicit review and confirmation moments.
Improve mobile first impressions by surfacing the main action immediately.
Create a developer-ready system that could scale across responsive layouts and bilingual content.
Research Approach
The team validated the concept through user interviews and usability testing. The research documented not just what users said they wanted, but how they currently behave when searching, comparing, validating prices, and deciding whether to trust a platform enough to book through it.
Two patterns stood out early. First, Booking.com was the dominant mental model for both participants, valued for habit, ease, and reliability. Second, openness to AI existed, but trust depended on repeated proof, transparent logic, and human approval before final commitment.
Key Insights
The research produced several important insights that shaped the redesign:
• Users need fast orientation before they need detail.
• Clear grouping matters more than adding more content blocks.
• Stronger hierarchy can improve usability without making the interface feel visually heavy.
• Users want AI support, but not full AI control over booking and payment.
• Trust is built through reviews, comparison, transparent pricing, and visible confirmation steps.
• Mobile hierarchy is critical, because the main action must be visible immediately.
These insights became the backbone of the final experience. They pushed the design toward a cleaner structure, simpler navigation, stronger visual hierarchy, and a more disciplined booking flow.
Design direction
The design direction moved away from traditional travel forms and toward a conversational interface that lets users describe their trip in natural language. This made the first step feel more human and less like a data-entry task, which was especially important for users who already know what kind of trip they want, but do not want to fight with filters from the start.
To support trust, I introduced transparent recommendation logic. Instead of showing an endless list of options with no explanation, the system was designed to help users understand why a specific flight or hotel appeared in the results. That reduced the “AI black box” feeling and gave the product a more credible, decision-support role.
I also prioritized the homepage hero and mobile layout so that the AI prompt would be visible immediately. The research showed that if the entry point is hidden, the entire concept feels weaker, even if the underlying product is strong.
What changed
One of the most important changes was introducing a mandatory Review & Confirm checkpoint. Testing showed that users felt anxious when AI appeared to take full control during checkout or disruption scenarios, so the flow needed a clear moment for human approval.
The visual hierarchy was also strengthened across the entire booking journey. Instead of allowing every step to compete for attention, the redesigned flow guides users more clearly through the process, reducing confusion and making the experience feel more controlled.
The mobile homepage was simplified so the chat prompt could act as the primary above-the-fold action. This was not a cosmetic change — it directly addressed the first-impression problem identified during research.
Lo-fi wireframes helped validate the structure before visual polish.
They let me test hierarchy, flow, and decision points early, before moving into hi-fi design.
Scalability & A Developer-Ready Handoff
Knowing this design needed immediate implementation, I built the UI with a strict developer handoff strategy. I utilized robust Figma Auto Layouts to ensure the interface was 100% responsive across desktop and mobile breakpoints. Crucially, I designed dynamic components capable of handling text expansion for the French translations without breaking the layout—a major usability issue in the client's original build. The final deliverable was a polished, scalable ecosystem shaped by teamwork and iterative design.
Outcomes & Reflections
Delivering the Kitravia project was a masterclass in balancing cutting-edge technology with the fundamental human need for trust and control. By shifting from a traditional, form-heavy interface to an intuitive, conversational AI model, we established a design foundation projected to reduce average booking time. Beyond the metrics, leading the B2C flow for a bilingual, global platform reinforced my core design philosophy. It proved that true innovation isn't just about implementing the latest AI features; it is about crafting digital journeys that translate complex, overwhelming data into an effortless, accessible, and deeply human experience.
Solution
Kitravia translated AI-assisted travel into a clearer, more trustworthy booking experience by combining conversational search, structured results, two booking flows, and a guided visa hub within one coherent system.[Załącznik]
Rather than automating decisions away from the user, the design used speed, transparency, and clear confirmation points to make complex travel planning feel simpler, more credible, and easier to act on
Link to the Prototype : https://www.figma.com/proto/iSugl1rJq8eWLnyUz1gAoS/Kitravia--Old-file?node-id=17-20165&t=b0qBCjaLL8wXSeMj-1&scaling=scale-down&content-scaling=fixed&page-id=7%3A7386&starting-point-node-id=17%3A20165
