A 0→1 MVP launch focused on clarity, structure, and helping users get started.

To make sure all the location actually exists we were adding location card via Google map search.

Locations could be added manually in the app or generated via chat. In the background, we queried Google Maps to validate each place before adding it to the itinerary.

Users can add places to a general list, immediately see them on the map, and then distribute them into days. We supported multiple ways to do this:

drag and drop

a utility action

searching directly within a day

so users could choose what felt most natural.

Based on post-launch learnings, the first priority was fixing the foundation. The card needed to be more reliable, easier to interact with, and scalable for future iterations. We also had a system constraint. Heavy Google Maps usage was driving costs, so part of this redesign was intentionally reducing how much data we pulled into the card while keeping the experience convenient. These requirements set the direction for the redesign.

I explored a few directions, including separating planning and traveling modes, and different ways of exposing actions and information density. Ultimately, we decided not to introduce modes. While conceptually clean, they added cognitive overhead and complexity that didn’t feel right for an MVP. Instead, we focused on improving the existing model.

From session recordings, I noticed users consistently adding flight, accommodation, and rental car details as notes. That was a clear signal that logistics mattered. We chose to start with flights because the data is structured, time-bound, and easier to integrate reliably than other logistics. This also created a foundation for future monetization through affiliate links, but the primary goal was improving clarity and preventing misuse of generic cards.