A place to chat and get assistance on ongoing and past bookings. Travel goes wrong mid-journey, so help has to work from an airport queue, not a desk.
The call I made
Design for the two highest-volume requests and let the bot own them completely, instead of building a general-purpose help surface that handles everything adequately.
The evidence came from reading real conversations with the CX team, not from a taxonomy of ticket types. Cancellations and reschedules dominated, so those got a flow measured in seconds and the agents got their judgment cases back.
I owned
I partnered on
Where the judgment was
Cleartrip's support relies primarily on chatbots to address enquiries quickly, which lets human agents focus on the more pressing matters. It works as an inbox for anyone who needs help with an upcoming or past booking: raise the question, and the bot either resolves it or connects you to the customer support team.
I served as the sole designer and established direct communication with both the CX team and the agents themselves — because they know what a query looks like before it has been tidied into a ticket category. The objective was to comprehensively address issues across all the possible scenarios, not just the tidy ones.
Present everywhere, simple enough to use under stress.
We aimed to make support visible across the entire application, integrating it from the profile page through to the My Trips section. This ensures travellers can reach support not only for a significant concern but for the smallest thing they run into anywhere in the journey — which is exactly when a small thing is cheapest to fix.
It was designed to be effortlessly interactive. The bot is crafted for simplicity in every interaction, so navigating and addressing a query is easy — which matters because the person using it is usually stressed, often on mobile data, sometimes with minutes to spare.
The chatbot worked best where customers wanted to cancel or reschedule a booking and get a refund. We read a lot of conversations about upcoming trips, worked out the main reasons people cancel, and built a simple flow around exactly those — so the traveller resolves it themselves in about ten seconds, and the agents get their time back for the cases that genuinely need judgement.
Open to full-time roles and freelance projects