Sity：运行在树莓派上的自托管 AI 助手，具备持久记忆与认知架构
Sity — AI assistant with persistent memory and its own cognitive architecture.
开发者 SrAlex16 用约 4 个月打造出自托管对话 AI 助手 Sity，后端、数据库、TTS/STT 全部本地运行在树莓派 4B（4GB RAM）上，语言模型则通过 API 调用 Claude Haiku。
I've been building Sity for the past ~4 months. It's a
self-hosted conversational AI that runs entirely on a
Raspberry Pi 4B (backend, database, TTS/STT — all local).
The language model is Claude Haiku via API, but everything
else is on-device.
What makes it different from a typical AI assistant:
It has a proper cognitive pipeline — not just a prompt with
personality instructions. Each conversation turn goes through:
Perception → emotional appraisal → goal tracking
Episodic and semantic memory that persists across sessions
A utility-based decision engine (10 possible actions)
Reflection and self-model that update over time
The relationship with the user evolves. It remembers things,
tracks expectations, and can take initiative when it detects
something worth following up on.
Self-hosted details:
Runs on Pi 4B (4GB RAM)
SQLite database, no external dependencies
Piper TTS + faster-whisper STT (both local)
Caddy reverse proxy + Cloudflare Tunnel for remote access
AGPL-3.0
Current status: public alpha, open for testing
🌐 https://sity.aletm.com/
💻 https://github.com/SrAlex16/sity
I'm particularly interested in feedback from people who've
built similar things — curious how the architecture compares
to other approaches.
— SrAlex16
来源：Google AI：DEV 作者专属（RSS） · dev.to