What I Built
YaadVault AI is an AI-powered digital memory vault designed to preserve family stories, personal experiences, and memories before they disappear.
We all have stories worth remembering. Our grandparents have childhood memories, our parents have life experiences, and our families have stories that are often passed down through conversations.
But what happens when those stories are forgotten?
That's the problem I wanted to solve.
YaadVault AI transforms spoken memories into searchable digital records using open-source AI.
Users can record a family member sharing a story, convert that recording into text, generate an AI-powered summary, and save it permanently in their personal memory vault.
The idea behind the project is simple:
Technology should not only help us build the future; it should also help us preserve our past.
Key Features
- 🎙️ Voice Memory Recorder: Record personal stories and family conversations.
- 📝 AI Transcription: Convert spoken memories into readable text.
- 🧠 Smart Summarization: Generate meaningful summaries of long recordings.
- 📅 Memory Timeline: Organize memories chronologically.
- 💬 Ask Your Memories: Ask questions about saved stories using AI.
- 🔍 Semantic Memory Search: Find memories based on their meaning, not just exact keywords.
- 🔒 Privacy-First Architecture: Keep personal memories and AI processing local.
Demo
📸 Screenshots:
The demo will showcase the complete journey: recording a memory, transcribing it, saving it, and asking questions about it
How I Built It
I wanted YaadVault AI to be more than another application that simply sends user data to a third-party AI API.
That's why I chose an open-source, local-first AI architecture.
Technology Stack
| Component | Technology |
|---|---|
| Frontend | Next.js |
| Styling | Tailwind CSS |
| Backend | Python + FastAPI |
| AI Model | Gemma |
| Local AI Runtime | Ollama |
| Speech Recognition | Whisper |
| Database | SQLite |
How the AI Works
- A user records a personal story.
- Whisper converts the audio into text.
- The transcribed text is passed to Gemma.
- Gemma generates a summary and extracts useful information.
- The memory is stored in the local database.
- Users can later ask questions about their saved memories.
For question answering, the application retrieves relevant saved memories and provides them as context to the language model.
This helps keep answers grounded in the user's actual recordings rather than generating fictional family history.
The goal is to make the entire experience work locally, without depending on a closed AI API for its core intelligence.
Why Does Open Innovation Matter?
Open innovation is especially important when building applications that deal with personal memories and sensitive information.
Using open-weight models and local inference makes several things possible:
1. Privacy and Control
Personal stories and family recordings are valuable. Local AI processing gives users more control over where their data goes.
2. Freedom to Experiment
Open-weight models allow developers to experiment with different models, prompts, and inference configurations without being locked into a single provider.
3. Accessibility
With tools like Ollama, developers can experiment with local AI without needing to build an entire AI infrastructure from scratch.
4. Community Collaboration
Open-source development allows other developers to improve the application, suggest features, fix bugs, and make memory preservation accessible to more people.
For me, open innovation means that powerful AI technology should not be limited to companies with expensive infrastructure.
It should be available to everyone who has an idea worth building.
The Spirit of It
Imagine sitting with your grandmother and asking her:
"What was your childhood like?"
She starts telling you stories about her village, her school days, festivals, and people she once knew.
You record that conversation.
Years later, someone in your family opens YaadVault AI and asks:
"What was Grandma's childhood like?"
Instead of searching through old notebooks or trying to remember conversations, they can explore those memories again.
That is what inspired YaadVault AI.
I believe memories are more than old photographs or recordings. They are pieces of our identity, our family history, and our connection to the people we love.
People may leave, but their stories don't have to disappear.
That's the little piece of the future I wanted to build. ❤️
Final Thoughts
Building YaadVault AI has been an opportunity to explore how open-source AI can solve a problem that is deeply personal.
The project is built around one simple belief:
Every memory deserves a place to live.
If you like this idea, feel free to explore the repository, contribute, or share your feedback.
Let's build technology that keeps our memories alive! 🕯️
Top comments (0)