Explor

Inspiration...

In an ever-connected world, young people are continuing to create spontaneous trips to locations around the world, with influences from social media and television. Due to the logistical, economic, and cultural shifts associated with GenZ, we saw an opportunity to streamline projects associated with their need to “see the world”. We wanted to create a solution that not only provides users with an innovative opportunity to find destinations, but also allows them to harness AI and swiping mechanisms to join travel groups quickly and informatively.

What it does...

What is Explor?

Think of it as Tinder for Travel. Explor is an all-in-one travel planning and social platform that lets users discover popular destinations, connect with like-minded friends, and plan trips together... all in one place. It removes the friction of travel planning by combining swipe-based discovery, real-time group coordination, and AI-powered assistance into a seamless mobile-first experience.

1. Intuitive Swipe-Based Decision Making

Users are presented with curated destination cards and simply swipe right to like or left to pass. This makes trip inspiration effortless and fun
Each card surfaces rich details at a glance: star rating, price tier, trip duration, region, and descriptive tags (e.g. Beach, Culture, Adventure)
Social proof is built in: if friends have already liked a destination, their names appear directly on the card ("Liked by Sarah & Marcus"), making group consensus organic and visible
Price and region filters let users narrow the deck to match their budget or travel preferences
Passed destinations aren't lost: they cycle back so users can reconsider previous passes

2. Group Chat Capabilities

Once friends match on destinations, users can spin up group chats directly within the app to start coordinating
Supports both 1-on-1 DMs and multi-person group chats, with custom group naming
Group membership is managed easily: creators can add or remove friends at any time via the Edit Group modal
Chats update in real-time via Firestore listeners, so messages and group changes are instantly reflected across all members' devices
The Gemini AI bot is automatically added to every group chat as a built-in travel assistant, ready to answer questions and help plan without leaving the conversation

3. Gemini API + Generative Note Taking

Every chat has an embedded AI travel assistant powered by Google's Gemini API, allowing users to ask destination-specific questions, get itinerary suggestions, visa info, packing tips, and more... all in context
The AI is chat-history aware, meaning it understands the ongoing conversation and can give contextually relevant responses rather than generic answers
Alongside the AI chat, a collaborative Planning Notepad is embedded within each conversation. Users can save ideas, bullet out itinerary items, and store important trip details mid-chat
The Notepad feeds context back into Gemini, so the AI can reference saved notes when making suggestions, creating a genuinely intelligent planning loop
The Gemini API is proxied server-side to keep API keys secure and off the client

4. Secure Matchmaking and Networking Menus

Google OAuth via Firebase Authentication provides secure, one-click login with no passwords to manage
A dedicated Friends system lets users send, accept, or deny friend requests. Only friends can be added to group chats, keeping the experience trusted and intentional
Mutual friend scoring surfaces the most relevant friend suggestions at the top. The more mutual connections, the higher the priority
Trip Match notifications alert users in real-time when a friend likes the same destination. Creating a natural moment to reach out and start planning
User profiles display a bio, liked destinations, group memberships, and mutual picks... clicking any friend's profile shows exactly where your travel tastes overlap, highlighted with a match badge
All data — messages, swipes, friendships, notifications — is stored and synced securely through Firebase Firestore with real-time listeners

How we built it...

React + Vite — frontend framework

Firebase Auth — Google OAuth authentication

Firestore — real-time database & data sync

Gemini API — AI travel assistant

Vercel — hosting & serverless API proxy

Tailwind CSS v4 — styling & design system

Mobbin + Canva — planning & UI design

Cursor — AI-assisted development

Foundation & Architecture

React + Vite frontend, Firebase (Firestore + Auth) backend — chosen for rapid prototyping speed
Tailwind CSS v4 with a custom design token system (brand colors, typography) in index.css

AI Assistance — Cursor!

Used Cursor as our AI coding assistant to accelerate development significantly
Handled boilerplate-heavy work: Firebase CRUD patterns, context providers, modal scaffolding, repetitive component structure
Let us focus creative energy on product decisions and UX rather than plumbing

Authentication & Live Users

Google OAuth via Firebase Auth for one-click login fully live and functional
seedUsers.js populates realistic fake users on first load so the social graph feels alive immediately
Real Firestore listeners mean friend requests, notifications, and messages update in real-time across sessions

Swiping Experience

Pointer-event drag system with rotation/opacity feedback and LIKE/NOPE overlays
Each card shows star rating, price tier, trip duration, and tags
Social proof layer: if friends already liked a destination, their names appear on the card — e.g. "Liked by Sarah & Marcus"
Price and region filters to narrow the deck

Social & Messaging

Full friend request system with mutual-friend-based suggestions
DMs and group chats: every group automatically includes the Gemini AI bot as a participant for in-chat travel planning. Available via Google Gemini API (gemini-2.5-flash model).
Planning Notepad embedded in chats: lets users collaboratively save ideas, itinerary notes, and suggestions mid-conversation

Profile Page

Summary stats: destinations liked, passed, and group chats joined
Editable bio, full liked destinations list with remove option
View any friend's profile to see their liked destinations: common picks are highlighted with a badge and sorted to the top

Judging Criteria...

Innovation & Creativity

Explor combines swipe-based destination discovery, social matchmaking, and AI-powered trip planning into a single cohesive product. While individual pieces exist elsewhere, the combination of Tinder-style travel swiping, friend match notifications, and a context-aware Gemini planning assistant in one app is genuinely novel. It reframes travel planning as a fun, social experience rather than a research chore.

Technical Complexity

The app integrates Firebase Auth, Firestore real-time listeners, the Gemini API, and a Vercel-proxied backend — all wired together in 36 hours. The social graph handles friend requests, mutual scoring, swipe syncing across users, and live trip-match notifications. The Gemini integration passes full chat history and notepad context on every request, requiring careful state management throughout.

Functionality & Execution

Explor is fully deployed and live — not a prototype. Auth, swiping, friend requests, real-time messaging, group chats, AI responses, and notifications all function end-to-end. It has been tested by real users with positive feedback and no reported critical bugs.

Impact & Usefulness

Group travel planning is genuinely fragmented — it typically spans multiple apps, spreadsheets, and message threads. Explor collapses that into one product: discovery, social matching, group chat, and AI planning in one place. Real users tested it during the hackathon and the feedback was immediate. This is a problem people recognize the moment they see the solution.

User Experience & Design

The UI is polished with smooth swipe animations, and a consistent warm brand system built with Tailwind v4 custom tokens. Modals, cards, and filters all follow a clean rounded design language. The UI/UX is intuitive with easy access and flow for older audiences as well.

Best Use of Vercel

Vercel serves the React/Vite frontend as a static build and handles a serverless API route (/api/gemini) that proxies all Gemini requests, keeping credentials out of the client bundle entirely. This serverless function pattern meant zero backend infrastructure to manage — no Express server, no separate hosting. Vercel's CI/CD pipeline also meant every push was instantly reflected in the live production URL, which was critical for iterating fast during the hackathon.

Best AI Application

The Gemini assistant is embedded directly inside group chats via a dedicated chat participant model, meaning AI responses share the same message thread as human coordination. Each API call passes a structured payload containing the full message history and current notepad entries, giving the model genuine conversational context rather than isolated prompts. This retrieval-augmented approach makes the AI useful for real trip planning rather than generic Q&A.

Best Use of Gemini API

Every Gemini request is constructed with multi-turn chat history and user-generated notepad content serialized into the prompt, enabling the model (Gemini 2.5 Flash) to reason across both conversational context and structured planning data simultaneously. This goes beyond single-turn prompting — the model functions as a stateful planning agent that gets more accurate as the conversation and notes accumulate. Responses cover itinerary generation, destination-specific logistics, and trip advice, all grounded in what the specific group has already discussed.

Best Design

The UI uses a cohesive design system with custom brand tokens, rounded card components, and smooth micro-interactions like swipe feedback and modal transitions. The swipe cards surface photography, ratings, pricing, and social proof in a hierarchy that feels immediately intuitive. The overall aesthetic is clean and consumer-grade, with careful attention to mobile layout, spacing, and touch targets throughout. We created our designs for our website and presentation with help from Canva, for creating our logo and slides, and Mobbin, for inspiration for our landing page and UI elements.

Challenges we ran into...

Gemini API & Deployment

Managing Gemini API credits was a real constraint initially during development and live demos with multiple users hitting the bot simultaneously, we had to be strategic. Fortunately, we were able to receive $300 in API credits (Thank you Google!)
Deploying to Vercel introduced challenges around proxying the Gemini API server-side securely, ensuring the bot worked reliably for all users rather than just locally

Instagram Integration — Paused

One of our most exciting planned features was a "Connect with Instagram" integration. DM our explor.with.ai account on Instagram and let AI put together the vacation for you.
We hit a hard wall: Meta requires a formal app review and webhook approval process that takes several days, which fell outside our hackathon window
This isn't abandoned, though! We've scoped it out and plan to ship it in the coming weeks as one of our first post-hackathon updates

Accomplishments that we're proud of...

Shipping a Live Product

Explor isn't a just mockup or MVP! It's a fully deployed, live web app hosted on Vercel, tested by real friends and family, and already receiving positive feedback
Knowing people actually used it and enjoyed it made the late nights worth it
We are still a long way from done... more updates to come!

Stepping Into New Territory

None of us had significant experience with Vercel, the Gemini API, or building polished front-end products at this scale. We figured it all out under pressure, in 36 hours
Every feature shipped was a genuine learning moment for us

Solving a Real Problem

The feedback we got made one thing clear: this fills a gap that actually exists
Group travel planning is notoriously messy amongst teens and young adults. It's scattered across iMessage threads, Google Docs, and endless DMs, and people responded to having it consolidated in one place

Building Efficiently with AI

Leveraging Cursor as a coding assistant let us move at a pace that simply wouldn't have been possible otherwise. Boilerplate code that would've taken hours got done in minutes, freeing us to focus on what made the product special
It was a masterclass in human + AI collaboration under real constraints

The Team

Brainstorming sessions, debugging rabbit holes, laughing at the mishaps and mistakes — the process was just as rewarding as the product!
Win or lose, this is another core memory for the books. As our senior year comes to an end we're genuinely grateful for each other, and this was time well spent. 🧡

What we learned...

Scoping is everything: with 36 hours on the clock, learning to cut boldly and ship a focused core product was more valuable than any individual technical skill we picked up
AI-assisted development is a superpower: Cursor changed how we approached problems; knowing you can generate boilerplate instantly means you think bigger and prototype faster
New tools aren't as scary as they look: Vercel, Gemini, Firebase, Tailwind v4; none of us were experts going in, and that taught us curiosity moves faster than comfort
Real feedback changes your perspective: hearing from actual testers <24 hours in, saying that this solves a problem they have, made us think less like “hackers” per say and more like product builders
Teams win: technical skills got us to a working app, but communication, trust, and being able to laugh mid-crisis got us to something we're actually proud of

What's next for Explor...

Instagram Integration

Did you really like a vacation reel on Instagram? DM our explor.with.ai account on Instagram and let AI put together the vacation for you. Nothing else needed!

Multi-Language Support

Destination cards, chat interfaces, and AI responses translated into Spanish, French, and Chinese. We hope to open Explor up to a genuinely global user base

Location-Based Discovery

Smart destination suggestions based on your current location or home city, surfacing relevant flights, travel times, and nearby hidden gems to make recommendations feel personal rather than generic

Booking Integration

Direct connections to platforms like Airbnb and Booking.com! So when a group agrees on a destination, the next step is one tap away. Closing the loop between "we should go here" and "we're booked"