My submission for the Hacktoberfest Open-Source AI Challenge Week 4: Ghost in the Machine.
💡 What I Built
Some memories are too precious to forget.
A person's voice, their favourite stories, small daily habits, and the little things they used to say can stay with us long after they are gone.
However, these memories are often scattered across photographs, personal notes, conversations, and our own thoughts.
I wanted to explore how AI could help people preserve and revisit these memories in a more personal way.
That idea became Echoes of the Lantern, a privacy-focused memory archive that uses browser-based AI to help people create remembrance stories and interact with saved memories.
Inspired by the visual idea of a remembrance altar, the project creates a calm digital space where users can record meaningful details about someone important to them.
The goal is simple:
Keep meaningful memories alive, one story at a time.
✨ Key Features
📝 Create a Memory
Users can create a personal memory by entering details such as:
A person's name.
Their relationship or connection.
A memorable story.
Favourite food or drink.
Favourite song or flower.
Small personal details that make the memory special.
These details become the starting point for creating a remembrance experience.
🏮 Remembrance Story Generator
Echoes of the Lantern uses AI to turn saved memory details into a remembrance-style story.
It helps organize personal information into a more readable and meaningful format.
The generated story can help users revisit important moments and remember the small details that make someone special.
AI-generated stories may sometimes add incorrect details, so users should review the output and ensure it accurately represents their memories.
📚 Personal Memory Archive
Users can view their saved memories in one organized place.
The archive provides a simple way to revisit previously created memories instead of searching through different notes and documents.
💬 Ask the Echo
This feature allows users to ask questions about their saved memories.
For example:
What was my grandfather's favourite morning routine?
Which food did he enjoy?
What small habit do I remember about him?
What advice did he give me?
The AI uses the saved memory as context to generate a response.
The feature is designed to make revisiting memories feel more interactive.
However, AI responses are not guaranteed to be accurate and should not be treated as a source of facts beyond the information saved by the user.
🔒 Privacy-Focused Experience
Privacy is an important part of Echoes of the Lantern.
The application is designed around browser-based memory storage and local AI inference.
Memories are stored in browser storage.
AI inference is designed to run on the user's device.
The application does not use an external AI inference server for memory generation.
No user account is required for the basic experience.
The hosted website still needs an internet connection to load. Model availability and initial model loading may also depend on the browser and network environment.
🎬 Project Demo
🌐 Live Website:
https://echoes-of-the-lantern.onrender.com/
🎥 Video Demo:
https://www.image2url.com/r2/default/videos/1793461841699-b23ef2a6-febb-4843-bfb7-336e2105e952.mp4
The demo introduces the project interface and shows how users can create a memory, generate a remembrance story, explore their personal archive, and interact with saved memories using Ask the Echo.
It also demonstrates the project's main idea: using AI to create a personal and privacy-focused digital space for meaningful memories.
💻 Source Code
GitHub Repository:
https://github.com/tanmaydave2007-netizen/echoes-of-the-lantern
The project is available for developers who want to explore its implementation, understand browser-based AI integration, or contribute improvements.
🛠️ How I Built It
Echoes of the Lantern uses a browser-based architecture with React and Vite.
The project uses the following technologies:
⚛️ React: Builds the interactive user interface.
⚡ Vite: Provides the development and production build environment.
🎨 CSS: Creates the visual design, layouts, colours, and responsive styling.
💻 JavaScript: Handles application logic and user interactions.
🤗 Hugging Face Transformers.js: Supports running AI models directly in the browser.
🤖 SmolLM2: The small language model used for local AI-powered memory interactions.
⚙️ Web Workers: Help keep AI processing separate from the main browser interface.
💾 Browser Storage: Stores users' memories in the browser.
🐙 Git and GitHub: Support source-code management and collaboration.
☁️ Render: Hosts the live web application.
The project explores how a small AI model can be integrated into a web application without depending on a remote AI inference server.
🔓 Why Open Innovation Matters
Open innovation makes AI development more accessible to independent developers and students.
Developers can explore available models, experiment with different architectures, study their behaviour, and build useful applications without training a large language model from scratch.
For Echoes of the Lantern, open-source AI technologies made it possible to explore a different approach to AI-powered applications.
🧠 AI on the User's Device
Many AI applications send user prompts to external servers.
Echoes of the Lantern explores browser-based local inference, where the AI model performs its generation work on the user's device rather than relying on a remote AI inference server.
This approach can help reduce the need to send personal memory content to an external AI inference service.
🔐 Privacy by Design
Personal memories can contain sensitive and meaningful information.
Keeping memory storage in the browser and using local inference helps explore a more privacy-focused architecture.
Users should still understand their browser storage, model-loading, and device-security limitations.
🧩 Accessible AI Development
Using a smaller model such as SmolLM2 allows developers to experiment with AI integration without necessarily requiring a powerful dedicated AI server.
Actual performance depends on the user's device, browser, and model-loading environment.
📚 Learning Through Open Technologies
Building this project helped me understand how frontend development, browser storage, Web Workers, and local AI inference can work together in one application.
Open technologies make it easier for students and independent developers to experiment, learn, and build new ideas.
👻 How It Supports the Ghost in the Machine Theme
The Ghost in the Machine theme inspired me to explore the connection between technology, memories, and AI.
A memory is not just a collection of facts.
It can include someone's habits, favourite things, conversations, stories, and the small details that make them unforgettable.
Echoes of the Lantern uses AI to create an interactive way to revisit these saved memories.
The AI acts as a digital companion that helps organize and explore the information users choose to preserve.
The project is not designed to recreate a person, imitate someone who has passed away, or claim that AI can bring memories back to life.
Instead, it explores how technology can help people preserve meaningful stories in a respectful and personal way.
A Simple Memory Workflow
Step 1: Open Echoes of the Lantern.
Step 2: Select Create a Memory.
Step 3: Enter meaningful details and personal stories.
Step 4: Generate a remembrance story using the AI.
Step 5: Save and revisit the memory in the archive.
Step 6: Use Ask the Echo to explore the saved details through questions.
This creates a simple digital process for preserving and revisiting important memories.
🚧 Current Limitations and Next Steps
Echoes of the Lantern is an evolving project, and there are several areas I want to improve.
My planned improvements include:
Improving AI response accuracy and reducing hallucinations.
Making generated remembrance stories more faithful to saved information.
Improving the Ask the Echo experience.
Adding better controls to edit and manage saved memories.
Improving backup and restore options.
Testing browser storage behaviour across different devices.
Improving model-loading and error-handling experiences.
Testing the experience across desktop and mobile browsers.
Improving accessibility and interface responsiveness.
Exploring additional small open-weight models.
Improving privacy documentation and user controls.
The current AI model may generate incorrect or unsupported details. Users should review generated content before saving or sharing it.
Local inference also depends on model availability and device capabilities.
📚 What I Learned
Building Echoes of the Lantern helped me gain practical experience with browser-based AI development.
I explored how to:
Build an interactive application using React and Vite.
Store information in browser storage.
Integrate a small language model using Transformers.js.
Use Web Workers for AI processing.
Connect AI functionality with a web interface.
Build a privacy-focused application architecture.
Manage source code using Git and GitHub.
Deploy a web application using Render.
Most importantly, I learned that building an AI application is not only about generating text.
A meaningful AI application also needs a clear purpose, a thoughtful user experience, responsible AI behaviour, privacy considerations, and honest communication about its limitations.
🎯 Final Thought
Some memories deserve a light that never goes out. 🏮
I built Echoes of the Lantern to explore how AI and open technologies can help people preserve meaningful memories in a personal and privacy-focused digital space.
The aim is not to replace human memories or emotions with artificial intelligence.
The aim is to use technology to help people organize, revisit, and appreciate the stories that matter to them.
Echoes of the Lantern — Keep the Memories Alive.
I welcome feedback, suggestions, and contributions from the open-source community!
Top comments (0)