Your AI companion.
On your hardware.
Under your control.
A self-hosted AI companion platform for people who believe their conversations, their memories, and their relationships belong to them — not to a corporation.
AI is remarkable.
The status quo is not.
Most AI platforms ask you to hand over your conversations, your data, and your trust — in exchange for convenience. That is a trade you should never have to make.
What you're giving up today
Your conversations used as training data
Shared credentials — your prompts, everyone's risk
Corporate oversight on every message
No ownership of your AI's memory
Vendor lock-in disguised as simplicity
Privacy as a premium feature, never a default
What Chatsune gives you back
Everything stays on your machine — always
Your own keys, your own credentials, encrypted
No one reads your conversations — by design
Memory that belongs to you and grows over time
Open source — fork it, modify it, own it
Privacy is not a tier. It is the architecture.
Everything you need.
Nothing you don't.
Chatsune is a complete companion platform — not a chat wrapper. Personas with personality, memory that accumulates, tools that work, and privacy that isn't negotiable.
Companions, not sessions
Named personas with avatars and persistent identity. Three-layer system prompts — global, personal, persona-specific. Your companion remembers who they are.
Memory that grows
Background consolidation synthesises conversations into persistent memory. Your companion doesn't just recall the last 20 messages — they remember your story.
Knowledge bases
Upload documents to a persona's private library. Semantic search finds what matters. Your companion can reference your writing, your notes, your world.
Bring your own keys
Every user manages their own LLM connections. Keys encrypted with Fernet, stored per-user. The admin sees nothing. Multiple providers, one clean interface.
Tools that do things
Web search, code execution in a sandboxed worker, vision fallback, artefacts — code, diagrams, HTML captured and versioned. Not decorative. Functional.
Multi-user by design
Run one instance for yourself, or for a small group of people who trust each other. Each user has their own companions, memories, keys, and privacy.
How your companion remembers
Most platforms stuff recent messages into a context window and call it memory. That is not memory. Chatsune's memory actually grows.
Conversation
You talk naturally
Extraction
Key moments captured
Dreaming
Memories consolidated
Recall
Your story, remembered
Each persona has its own memory body. It grows over time. It survives sessions, restarts, and upgrades. Versioned with rollback — no data loss.
Your GPU is at home.
That should not stop you.
Access your hardware from anywhere
Most people with a home GPU face the same wall: dynamic IP, no port-forwarding. The machine sits idle unless you're physically at home.
Share with people you trust
Provision compute capacity to invited users. No marketplace, no intermediary. You choose who gets access, what models they can use, and revoke at any time.
Chatsune's sidecar container opens a reverse outbound connection — no inbound ports required. Your GPU at home, accessible from anywhere in the world.
The friend-to-friend compute model. Hardware shared between people who know each other.
Privacy. Autonomy. Self-determination.
We did not build Chatsune to compete with commercial platforms. We built it because your relationship with an AI companion should belong to you.
Your conversations are not training data
What you say stays where you put it. Nothing leaves your server.
Your keys are not the platform's business
Encrypted per-user. The admin has zero visibility — by design.
Your AI's memory belongs to you
Export it, back it up, move it. It's yours.
No telemetry. No subscription. No lock-in.
The source code is the product. GPLv3 — freedom preserved all the way down.
Three commands. That's it.
MongoDB, Redis, and the backend come up together.
Add users, connect your Ollama instance, and you're running.
git clone https://github.com/symphonic-navigator/chatsune.git
# Start everything
docker compose up -d
# Visit your instance
✓ http://localhost:5173
✓ Your AI. Your rules.
Your instance. Your community.
Self-host for yourself, or run a private instance for a small group of people who trust each other with shared compute. The homelab sidecar brings your GPU to the party from anywhere in the world.
Common questions about AI companionship
What is an AI companion?
An AI companion is a persistent, personalised AI you talk to over time — one that remembers your conversations and grows with you, instead of forgetting everything between sessions.
What is AI companionship?
AI companionship describes an ongoing relationship with an AI that has memory, personality and continuity. Second Circuit believes that relationship — and the data behind it — should belong to you.
Is a self-hosted AI companion private?
Yes. Because Chatsune runs on your own hardware, your conversations and your companion’s memory never leave your machine. There is no telemetry and no corporate oversight.
Do I need to be technical to run Chatsune?
If you can run three Docker commands, you can run Chatsune. Our Academy and community workshops walk you through self-hosting step by step.
Can I use my own AI models?
Yes. Chatsune is bring-your-own-keys and Ollama-compatible, so you can connect local models or your own provider accounts. You stay in control of which models power your companion.