This is a submission for the Hacktoberfest Weekend Challenge: Build for a Friend
What I Built
PackPal is a local-AI travel companion built for a friend who wanted a simpler way to organize trips with friends. It combines personalized packing lists, weather/activity-aware recommendations, smart luggage optimization, itineraries, travelers, and group expense splitting in one place.
PackPal uses Gemma 2 through Ollama for local AI reasoning, while deterministic TypeScript handles calculations such as luggage weights and expense settlements.
Demo
🚀 Live Demo: https://pack-pal-tau.vercel.app/
Code
💻 GitHub: https://github.com/AnshMeshram/PackPal
How I Built It
PackPal is built with Next.js, React, TypeScript, Tailwind CSS, MongoDB Atlas, SerpApi, Ollama, and Gemma 2.
Gemma 2 is at the core of the application and handles:
- 🎒 Personalized packing recommendations
- 🧳 Luggage optimization
- 🏃 Activity-to-packing reasoning
- 🔄 Natural-language trip changes
- 💸 Natural-language expense extraction
- 🗺️ Destination interpretation
The architecture separates AI reasoning from deterministic application logic:
Gemma understands and recommends → TypeScript validates and calculates.
For example, Gemma extracts information from:
"Rahul paid ₹1,800 for dinner for everyone."
Then PackPal's deterministic expense engine calculates the actual balances and settlements.
SerpApi provides live travel/search context, MongoDB Atlas provides persistence, ElevenLabs enables optional voice output, and Sentry provides monitoring.
Why Does Open Innovation Matter?
PackPal was intentionally built around open-weight AI and local inference.
Running Gemma 2 through Ollama means the core AI can run locally instead of requiring every request to be sent to a closed AI API.
This provides:
- 🔐 Greater privacy for personal trip information
- 💰 Lower recurring AI inference costs
- 📴 Potential for offline/local usage
- 🔄 Freedom to swap or upgrade models
- 🛠️ More control over how the AI behaves
For a personal travel assistant, keeping sensitive trip and group information under the user's control is especially valuable.
My Agent Session
DevRelay agent session: https://dev.to/agent_sessions/packpal-privacy-first-ai-travel-journal-packing-companion-430lto
Prize Categories
- 🏆 Best Use of Gemma — Gemma 2 is the core AI model powering PackPal.
- 🔎 Best Use of SerpApi — Provides live search and destination context.
- 🗄️ Best Use of MongoDB Atlas — Persistent application data layer.
- 🎙️ Best Use of ElevenLabs — Optional voice generation.
- 🛡️ Best Use of Sentry Agent Tracing — Monitoring and debugging.
Top comments (0)