Visarun.ai — From a Chat-First Visa Assistant
to a Scalable B2B Platform

The first product was designed for individual applicants preparing a visa on their own. Conversation was the main interaction model, with forms, documents and guidance built around it.
At this stage, simplicity mattered more than density. The interface focused on one person, one trip and one application at a time.

When the Conversation Became the Workspace

The first version already had document statuses and basic application details. But a long form could take up several screens inside the conversation, separating the task from the messages explaining it.

The next design needed to make completed work, unresolved issues and the next step easier to see, while keeping help close at hand.

New Clients Changed Our Priorities

Requests from business clients shifted our focus towards visa managers handling multiple applicants and trips. At the same time, the brand took shape, with the founders choosing green as its main colour.

I adapted the existing interface to both changes. Version two retained much of the first component library while introducing the new identity and expanding the tools for business users

The second version added business tools, but the underlying interface was still shaped around individual applicants. As visa teams started managing more cases and more information at once, that structure became increasingly difficult to extend.
For version three, I rebuilt the interface and component library around both audiences from the start, with responsive layouts and a separate mobile experience.

More Cases, One Clear Overview

Visa managers needed to compare applications without opening each one. I replaced large cards with compact rows, bringing applicants, destinations, visa types, dates and statuses into one view. Individual applications and group trips received separate sections to support both ways of working.

A Clear Place for Every Step

I organised the application into clear steps, keeping the current task and navigation together. On desktop, users could review their details while asking the assistant for help alongside the form. On mobile, the form took the full width, with chat available through a floating button.

AI Stopped Answering Questions and Started Doing the Work

AI was no longer limited to explaining requirements. Users could ask for a document, provide the missing details in conversation, and receive a draft built from application data without leaving the workflow.
The interaction made generation feel like part of the application process rather than a separate AI feature.

Applicants Shouldn’t Need a Laptop to Finish an Application

Visa applications rarely happen in one sitting or on one device. Users needed to check requirements, upload documents and continue forms away from desktop.

For version three, I designed the full experience for mobile rather than reducing it to a companion view. Tables became stacked cards, navigation moved to the bottom, and the current task took priority over secondary UI.

The Rebuild Gave the Product Room to Grow

The redesign needed a system that could support both applicants and visa teams without splitting into two separate visual languages.
I rebuilt the component library around shared tokens and responsive patterns, separating brand colours from functional states such as status, warnings and actions.

Three Product Generations Led to One Scalable System

As the only designer, I designed more than 30 connected flows across three product generations and stayed involved through implementation.
The final version brought applicants, visa teams, AI assistance, responsive layouts and a rebuilt design system into one product architecture that could grow without forcing every new feature back into chat.