Triply - Plan Group Trips Together

Rooms
Add and track bookings
Expense tracker and splits
Chat with friends and AI
Home page
Shared gallery

Inspiration

Every group trip starts the same way. A WhatsApp thread that spirals into 400 messages, three Google Docs nobody updates, a Splitwise link someone forgets, and a flight confirmation buried in someone's inbox.

We have lived this enough times to know the problem is never the trip. It is the tools.

Triply started as one question: what if your entire trip lived in one place, and your group actually had a shared brain for planning it?

What it does

Triply is a room-based group travel app. A user creates a room, shares an invite link, and the whole group is in. From there, everything lives in one shared space:

AI Agent - tag @Triply anywhere in the group chat to get destination suggestions, itineraries, cost estimates, and travel advice, with full context of everything the group has discussed
Live group chat with polls, destination voting, and date availability tools built in
Budget tracker to log shared expenses, split them, and settle up without switching apps
Travel manager to add flights, hotels, and bookings to a shared timeline with automatic reminders
Personal vault to store passports, visas, and documents, handy throughout the trip
Trip gallery - one drop zone for everyone's photos and videos, with reactions and downloads, no chasing people on WhatsApp

How we built it

Frontend: React with Tailwind CSS, built and iterated rapidly on Lovable
Backend: Supabase for auth, database, real-time subscriptions, and file storage
AI agent: Claude API (claude-sonnet-4-6) with the full trip chat history passed as context on every call
Architecture: Each room supports multiple trips. Every section (chat, budget, travel, gallery) scopes its data to the active trip

The AI agent is positioned as a trip planning specialist who knows what the group has already discussed, so answers are contextual rather than generic.

Challenges we ran into

Managing context for the AI agent was the hardest problem. The full chat history needed to be passed on every @Triply call, but as conversations grew, staying within token limits while keeping responses fast required careful trimming logic.

Real-time sync across four simultaneous sections also pushed Supabase Realtime harder than expected. Ensuring that an expense added by one member reflected instantly for everyone else, without race conditions, took iteration.

If \( n \) is the number of active members and \( m \) is the number of concurrent edits, synchronization complexity scales as \( O(n \times m) \), which became noticeable with larger groups during testing.

Accomplishments that we're proud of

The @Triply agent feels genuinely embedded in the conversation, not bolted on. It references earlier messages, remembers the budget the group mentioned, and gives answers specific to the group's actual plan.
Building a working, multi-section real-time product in under 48 hours using Lovable.
The shared gallery solved a real post-trip problem: photos scattered across five people's phones, now in one place.

What we learned

AI works best when it has context. A generic travel chatbot is forgettable. One that says "based on the budget Rahul mentioned earlier, here are three options" is actually useful.
Depth beats breadth in a hackathon. Three sections that work flawlessly impress more than seven that half-work.
Lovable is genuinely fast for product people who know exactly what they want to build.

What's next for Triply - Plan Group Trips Together

Reels integration - surface high-rated Instagram Reels and YouTube Shorts for any destination the agent recommends
UPI nudge links - one tap opens UPI with the exact settlement amount pre-filled
Map view - all pinned places on one map, grouped by day and area
Packing checklist - auto-generated per destination, shareable to the group
Smart document vault - expiry alerts for passports and visas before they sneak up on you

The long term goal: Triply becomes the operating system for group travel, from the first "let's go somewhere" to the last rupee settled.