Skip to main content

AETHON

A self-hosted, provider-agnostic personal AI assistant — Web UI, CLI, and messaging bots, with memory, multi-agent specialists, SOPs, a scheduler, telemetry, and a live dashboard. You run it; you choose the backend.

PyPIPythonLicenseStrands
pip install aethon-ai
aethon init # pick a provider, paste a key (or go local)
aethon start # → http://127.0.0.1:18790

One assistant, fully yours

AETHON is a single Python package that gives one memory-backed assistant every entry point you need — and never takes your data or your model choice out of your hands.

🔌

Bring your own provider

Defaults to OpenAI, or point it at any OpenAI-compatible endpoint (vLLM, LM Studio, LocalAI), the Anthropic API, a fully-local Ollama model — plus Bedrock, Gemini, LiteLLM, and Mistral. Flip one line of config.

💬

Every channel, one package

Terminal CLI, a browser WebChat UI, and Telegram / Discord / Slack bots (WhatsApp experimental) — all shipped in a single install, talking to one persistent assistant.

🧠

Memory & specialists

Long-term SQLite vector memory, plus Coder / Researcher / Analyst / Planner sub-agents the orchestrator delegates to. It remembers what matters and farms out the hard parts.

⚙️

SOPs & scheduler

Reusable slash-invoked workflows (SOPs) plus a cron scheduler that runs them on a timetable and delivers results to any channel.

📊

Live dashboard

Watch sessions, memory, telemetry, agents, and SOPs in real time — with session recording and replay, and a pixel-agent "live company" view.

🔒

Local-first & private

Services bind to 127.0.0.1 by default and your data lives under ~/.aethon. Workspace sandbox, blocked-command filtering, approval hooks, and a memory guard keep you in control.