Echoes — The Living Memory Archive

Inspiration

We were inspired by the idea of preserving personal memories while ensuring AI never invents the truth. Echoes aims to make life stories searchable, connected, and meaningful.

What it does

Echoes connects personal memories through a Memory Map and enables AI-powered conversations. Its Truth Boundary distinguishes source-backed facts, interpretations, and unknowns, supported by provenance tracking, integrity seals, and user-controlled deletion.

How we built it

We used React, TypeScript, Vite, and Tailwind CSS for the frontend; Python and Flask for the backend; Supabase for authentication and storage; and Groq's GPT-OSS-120B for AI-powered memory analysis and conversations.

Challenges we ran into

Our main challenges were preserving the distinction between facts and assumptions, connecting related memories, implementing secure memory handling, and ensuring AI responses remain grounded in the available evidence.

Accomplishments that we're proud of

We built a memory archive concept that combines AI conversations, a visual Memory Map, provenance, cryptographic integrity verification, and a Truth Boundary that acknowledges uncertainty instead of guessing.

What we learned

We learned how to integrate AI with a structured memory system, apply encryption and integrity checks, manage user-specific data, and design AI interactions that communicate uncertainty transparently.

What's next for Echoes — The Living Memory Archive

We aim to enhance voice-based memory capture, improve memory connections, support richer multimedia archives, and strengthen evidence-grounded conversations while giving users greater control over their personal memories.

Built With

• aes-gcm
• flask
• gpt-oss-120b
• groq
• lucide
• numpy
• openai-python-client
• oxlint
• postgresql
• pytest
• python
• react
• react-flow
• react-router
• scipy
• sha-256
• supabase
• supabase-auth
• tailwind-css
• typescript
• vite
• vitest
• web-crypto-api