Key points:
Fully local: no online accounts or cloud dependency
Journaling: supports structured personal reflection alongside work/project logs
Voice-first: integrates Whisper for transcription, piper for TTS
Searchable memory: turns scattered notes into structured, queryable knowledge
Use cases: journaling, project tracking, meeting decisions, research progress, client preferences etc
Stack: FastAPI backend, React frontend, SQLite, Ollama for local LLMs, BGE embeddings, Whisper, Piper TTS
The goal of Echo is to be a privacy-first alternative to cloud AI assistants — one that combines journaling and productivity without giving up control of your data.
Repo: https://github.com/29sayantanc/Echo
Open for suggestions and feedback!
sayantanc•5mo ago
Echo isn’t just for journaling — it also supports multiple profiles for multiple use cases and each profile has separate DB so no data contamination. You can create a personal profile a wrok profile etc.
Everything runs locally: Whisper for transcription, BGE for embeddings, Ollama for LLMs (e.g. Qwen, Mistral), and Piper TTS for speech. No online accounts or cloud dependencies — all data stays in a local SQLite DB. It doesn't even need internet to run!
Tech stack is FastAPI backend + React frontend. It should run fine on most laptops; model choice depends on your hardware.
Roadmap is intentionally open — I’d love feedback on the memory/search design, UX, and integrations.
Repo: github.com/29sayantanc/Echo
Happy to answer any questions or hear what features would make Echo most useful for you.