(As at 1 day before hackathon end date, seemed this was tested. However at that point the project was not yet finalized. Please ensure to test again as I am wrapping up before deadline date just submitted early to get all the details in while still improving app. Thanks!)

Alonis

Alonis is a data-enriched, AI-powered personal space to reflect, discover, and understand who you are.

Inspiration

In a world of constant external noise, where is the space for internal reflection? We share our lives on social media, but our deepest thoughts, goals, and patterns often remain unexplored. Alonis was inspired by the need for a truly private, intelligent digital confidant.

I imagined a "second brain" that you could talk to as naturally as a friend—an AI that doesn't just respond, but listens, remembers, and helps you connect the dots of your own life. Moving beyond buttons and forms to create a space where self-discovery is driven by natural, reflective conversation.

What it Does

Alonis provides a suite of tools designed to help you build a deeper relationship with yourself, powered by a single, context-aware AI.

• 🧠 Uncover Personal Insights: Engage in natural, text-based conversations for assessments like the Big Five Personality Test or a Mental Health Check-in. Alonis tracks your evolution over time, showing you how you're growing.
• ✍️ Capture Your Thoughts & Track Ambitions: A seamless, private space to jot down fleeting thoughts, detailed notes, and ambitious life goals. You can track your progress and mark goals as "achieved" to celebrate your milestones.
• ✨ A Hyper-Personalized Discovery Engine: Alonis learns from your assessments, notes, and chats to provide tailored recommendations for books, movies, music, and news from the Qloo API, helping you discover new interests that resonate with your unique profile.

How We Built It

Alonis is a full-stack application that integrates multiple AI and data technologies with a modern, dynamic frontend.

Tech Stack:

• Frontend: A fully responsive web application built with Flask, HTML5, CSS3, and modern JavaScript (ES6+). It implements asynchronous data loading (the fetch API) across the app for a fast, single-page application feel without a traditional frontend framework.
• Backend Core: Powered by OpenAI's GPT models for natural language understanding and conversation.
• Context & Memory: LangChain is used to create a persistent, context-aware agent that remembers user interactions across all features, from chats to notes.
• ML Models: Custom machine learning models for specialized assessments, complementing the Big Five Personality Model framework.
• Recommendation Engine: The Qloo API provides a rich, multi-category recommendation backend (movies, books, music, etc.).

Challenges We Ran Into

• Fine-tuning the LLM: A significant challenge was ensuring the LLM could maintain a natural, supportive conversational flow while reliably extracting structured data needed for our machine learning models.
• Meaningful Recommendations: Moving beyond generic suggestions to provide truly hyper-personalized recommendations required deeply integrating the user's personal context (from chats and notes) with the powerful Qloo API.
• Building a Dynamic Frontend: Creating a highly interactive and responsive UI from scratch without a major framework like React or Vue was a challenge. I focused on modern CSS and asynchronous JavaScript to build an app that feels like a fast, real-time conversation.

Accomplishments That We're Proud Of

I'm incredibly proud of developing a full-stack application that seamlessly integrates a complex backend (LLM, ML, APIs) with a polished, modern, and fully responsive frontend.

A major personal accomplishment was bringing this project's UI to life. As a backend and ML developer accustomed to prototyping with Streamlit, building a complete, dynamic, and responsive user interface from the ground up with Flask and modern JavaScript was a challenging and incredibly rewarding journey. This project represents the leap from a functional ML prototype to a full-fledged, user-centric application.

Most of all, I'm proud of creating a truly interactive and "live" user experience. By using asynchronous JavaScript throughout the app, we moved beyond static pages to build an application that updates instantly, making the entire experience feel like a single, cohesive conversation with a personal AI.

What We Learned

This project was a deep dive into building modern web applications. Key learnings include:

• The power of template inheritance (Jinja2) and component-based CSS for building a maintainable and scalable frontend.
• The importance of asynchronous design (fetch API) for creating a high-performance user experience that feels fast and fluid.

• The value of high-quality, specialized APIs like Qloo to rapidly build rich, data-driven features.

• For a backend developer, this was a hands-on masterclass in modern frontend development. I learned how to structure a responsive UI with advanced CSS, manage application state with asynchronous JavaScript, and create the seamless, non-reloading experience users expect—a significant step up from the rapid prototyping capabilities of tools like Streamlit.

What's Next for Alonis

The potential for Alonis is vast. Future avenues for development include:

1. Adding more interactive assessment environments (e.g., career aptitude, learning styles).
2. Expanding the recommendation categories to include podcasts, travel destinations, and local events.
3. Integrating locational context to provide timely, relevant suggestions based on the user's environment.

Built With

• css
• flask
• html
• javascript
• langchain
• openai
• qloo