This is a submission for the Hacktoberfest Open-Source AI Challenge Week 1: Touch Grass
What I Built
TerraMind is an offline-first AI nature companion built for the Hacktoberfest 2026 — Touch Grass Challenge. 🌿
We spend so much time staring at screens that we often forget to explore the world around us. I built TerraMind to turn AI from something that keeps us online into something that encourages us to step outside.
What does TerraMind do?
🌱 AI-Powered Nature Missions: Generate personalized outdoor activities based on your available time, energy level, interests, and location hints. Whether you have 10 minutes for a mindful walk or a few hours for birdwatching, TerraMind helps you find something meaningful to do.
📓 Private Nature Journal: Record observations, thoughts, and discoveries from your outdoor adventures. Keep track of what you see, hear, and experience in nature.
🧭 Distraction-Free Mode: Follow your mission checklist, focus on the present moment, and put your phone away instead of endlessly scrolling.
🔒 Offline-First & Privacy-Focused: TerraMind uses locally running open-weight AI models through Ollama for mission generation. Journal entries, mission progress, and settings are stored locally in the browser using IndexedDB, helping keep personal data on your device.
🌲 Built for Real-World Exploration: From noticing birds and identifying interesting natural patterns to practicing mindfulness and observing trees, every mission is designed to encourage curiosity about the world outside your screen.
Who is it for?
TerraMind is for students, nature lovers, people who spend too much time online, and anyone who wants to build a healthier relationship with technology.
The idea is simple: use AI less as a destination and more as a starting point for real-world experiences.
Instead of asking AI to keep you entertained for hours, let it give you a reason to close your laptop, step outside, and discover something new.
🌍 Touch grass. Notice more. Live a little offline.
Demo
🌐 Live Website: https://terra-mind-blond.vercel.app/
Code
💻 GitHub Repository: https://github.com/azmolwasimhussain-ops/TerraMind
How I Built It
TerraMind is built around Ollama and open-weight AI models, with one core goal: use AI to help people spend less time on their screens and more time exploring the real world.
🧠 Open-Source AI at the Core
I integrated Ollama to run AI models locally instead of relying entirely on paid, cloud-hosted AI APIs. TerraMind supports models such as Llama 3, Gemma 2, and Qwen, allowing users to choose a model based on their hardware and preferences.
The AI uses a user's available time, energy level, interests, and preferred outdoor location to generate personalized nature missions, practical instructions, safety tips, and interactive checklists.
🌿 Built for Offline-First Experiences
The application is designed to keep personal nature observations on the user's device.
- Local AI inference: Ollama can generate missions using locally running models.
- Offline fallback: A built-in botanical mission engine provides an alternative when local AI generation is unavailable.
- Private storage: IndexedDB stores missions, checklist progress, journal entries, and application settings locally in the browser.
- Progressive Web App: PWA support helps make the experience accessible on supported devices, including when offline assets are available.
- Resilient error handling: Model connection failures and malformed AI responses are handled with clearer diagnostics and fallback options.
⚙️ The Technical Approach
I built the application with a component-based frontend, API routes for AI integration, local database persistence, and a nature-inspired responsive design system. I also focused on keyboard accessibility, mobile usability, mission history, and a distraction-free mode.
Rather than making AI the destination, TerraMind uses AI as a starting point: generate a mission, put the phone away, explore your surroundings, and record what you discover.
The idea behind the project is simple: AI should not always keep us connected to a screen. Sometimes, it should help us disconnect.
Why Does Open Innovation Matter?
Open innovation matters to TerraMind because AI should be accessible, private, and adaptable—not locked behind expensive APIs or proprietary platforms.
By building around Ollama and open-weight models, I could explore local AI inference and make AI-generated nature missions possible without requiring every generation to go through a cloud-based AI provider.
Here is what open innovation made possible:
- 🔓 Freedom to experiment: I can try different models, compare their outputs, and choose one that works best for my use case and hardware.
- 🔒 Privacy by design: With local inference and browser-based IndexedDB storage, personal preferences, missions, and journal entries can remain on the user's device when the application is configured and used locally.
- 🌱 Accessibility: Users can experiment with AI without needing a paid, per-request API subscription, although running models locally still requires suitable hardware and storage.
- 📴 Offline potential: Local models and a built-in mission fallback make it possible to keep the core experience useful when cloud services or internet connectivity are unavailable.
- 🛠️ Community-driven innovation: Open tools let developers inspect, adapt, improve, and build on existing technology instead of starting from scratch or depending on a single vendor.
A closed API could still generate personalized missions, but it would typically introduce an external service dependency and potentially usage costs. Open tools gave me more control over how TerraMind runs and how personal data is handled.
For me, open innovation is about more than using free tools. It is about giving people control over their technology. TerraMind uses that freedom to turn AI into a bridge between the digital world and the natural one—helping people put their phones away, explore their surroundings, and reconnect with nature.
TerraMind’s Open-Source AI Stack
TerraMind is built around open-weight AI and local-first technologies that make personalized outdoor exploration more accessible and privacy-conscious.
🧠 Ollama — Local AI Runtime
Ollama runs compatible open-weight models locally, allowing TerraMind to generate nature missions without requiring a paid cloud AI API for each request.
🌱 Open-Weight Models
TerraMind supports locally installed models such as Llama 3, Gemma 2, and Qwen. Users can experiment with different models and choose one that fits their hardware and needs.
⚙️ Application Architecture
- Next.js and TypeScript: Application structure, user interface, and API integration.
- React and Tailwind CSS: Responsive, reusable UI components and styling.
- Ollama API: Local model discovery, connection checks, and mission generation.
- IndexedDB: Local persistence for missions, journal entries, checklist progress, and settings.
- Progressive Web App technologies: Offline access to supported cached assets and saved content.
🔒 Local-First by Design
The combination of local model inference and browser-based storage gives users more control over their data. Personal journal entries and saved missions can remain on their device, while the local AI generates personalized activities when Ollama is available.
When the model cannot be reached, TerraMind can use its built-in botanical mission fallback instead of pretending that an AI request succeeded.
The key idea: Open-weight AI provides the flexibility, while local-first storage and offline-friendly design help turn that flexibility into a practical nature companion.
Development Process
I used AI-assisted development to shape TerraMind's concept, build its interface, and develop its local-first AI experience. Throughout the process, I focused on a simple goal: make technology a starting point for outdoor exploration rather than another reason to stay in front of a screen.
Prize Categories
- Best Use of Gemma — TerraMind uses Google's open-weight Gemma model through Ollama for local AI-powered nature mission generation.
Top comments (0)