Start here
New to the platform? Read what SecondStack is and how it is self-hosted on your own infrastructure.
SecondStack Documentation
Own your AI infrastructure. A self-hosted, multi-model LLM platform for deploying, managing, and scaling conversational AI across your organization — chat, agents, gateway, guardrails, and budgets, all on your own infrastructure.
SecondStack is a self-hosted, multi-model LLM platform. It gives your organization one place to run conversational AI — a chat app across Claude, GPT, Gemini, and self-hosted open-weight models; an LLM gateway with virtual keys, guardrails, and budgets; an agentic Cowork Agent; Knowledge Collections for RAG; and admin plus user dashboards. Everything runs on infrastructure you control, so conversations, documents, and usage data never leave your environment.
Explore the docs
Section titled “Explore the docs”Chat & agents
Explore the Chat App across every provider and the Cowork Agent that runs Claude Code in sandboxed containers.
Gateway & controls
Centralize traffic through API Routing, and manage models, providers, and policy from ControlTower.
Cost & knowledge
Keep spend predictable with Cost Management, and ground answers in your own docs with Knowledge & RAG.
What the platform includes
Section titled “What the platform includes”- Chat App — multi-model chat with image generation, voice input, file attachments, RAG, and agents.
- Cost Management — per-user, per-team, and per-API-key budgets with threshold alerts and usage analytics.
- Cowork Agent — an agentic mode that runs Claude Code in per-thread Docker sandboxes with built-in Skills.
- Knowledge & RAG — named, shareable document sets auto-indexed for retrieval inside any chat.
- API Routing — a LiteLLM gateway with virtual API keys, guardrails, and spend observability.
- ControlTower — the admin dashboard for models, providers, prompts, guardrails, and access control.
- User Dashboard — self-service API keys, personal budgets, and per-request usage logs.