The problem
Group travel planning usually gets fragmented across message threads, spreadsheets, saved links, and several booking sites. The hard part is not finding one more recommendation; it is keeping preferences, budgets, timing, and decisions aligned across a group.
The approach
william.travel treats a trip as a shared workspace. Travelers can shape an itinerary together, keep practical details near the plan, and use AI as an assistant for drafting options rather than as an opaque decision maker.
The product direction emphasizes:
- recommendations grounded in the trip’s constraints;
- an itinerary that remains easy to edit by hand;
- collaborative decisions and budget awareness; and
- clear paths to maps and live travel data.
Architecture
The interface is built with React and TypeScript. A Node.js service owns application workflows, PostgreSQL stores structured trip data, and AWS provides the deployment foundation. External travel and AI services are kept behind server-side boundaries so their credentials and failure modes do not leak into the client.
What I am exploring next
The next useful questions are less about adding more generated text and more about trust: showing why a recommendation fits, making conflicts visible, and helping a group converge on a plan without losing individual preferences.