TravelMind
Planning trips effortlessly
The problem TravelMind solves
TLDR; TravelMind makes trip planning smarter, faster, and more personalized — turning what used to be a tedious process into a smooth, guided experience.
Planning a trip often means jumping between multiple tools — checking flights, comparing hotels, managing budgets, and trying to organize it all into one plan. It’s time-consuming, messy, and often confusing to get a clear, optimized travel plan that fits your needs and budget.
TravelMind solves this by acting as a unified travel assistant powered by the Model Context Protocol (MCP). MCP plays a key role by connecting different tools — such as route finders, weather checkers, event planners, and budget analyzers — into one coordinated workflow. This orchestration allows TravelMind to automatically gather data from multiple sources, analyze it, and present meaningful results in one place.
Through this setup, users can:
- Plan trips effortlessly with AI-generated day-by-day itineraries.
- Compare different travel options (flights, trains, stays, routes) in one view.
- Optimize their budget by balancing costs across transport, accommodation, and activities.
Challenges we ran into
-
Setting up an MCP (Model Context Protocol) client–server–host system turned out to be quite challenging, especially when using the TypeScript SDK. The biggest hurdle was that there wasn’t any ready-made host implementation available. Because of that, the client and server couldn’t communicate in a real orchestration setup out of the box. We had to build and simulate our own orchestration layer from scratch — handling things like data flow, request management, and response syncing between tools and models manually.
-
Another challenge came while integrating Gemini. We needed to design clean templates and structured prompts so that it could understand and respond properly. Getting the right schema, argument filling, and output format took a lot of trial and error since there weren’t many examples or guidelines available. Making Gemini produce consistent, well-structured responses was harder than expected