About Automatos AI

What Automatos AI is and how it works.

Automatos AI is an open-source multi-agent orchestration platform. It lets you create specialised AI agents, give them tools and knowledge, and coordinate them to handle complex tasks autonomously.

Core ideas

Agents are workers, not chatbots. Each agent has a specific role (Code Reviewer, Sentinel, QA Engineer), its own model configuration, persona, and set of tools. They don't just answer questions — they do work.

The platform routes, you don't have to. A multi-tier Universal Router analyses every message and sends it to the best agent automatically. You can override routing or let the system learn from corrections.

Knowledge powers everything. Upload documents, sync cloud folders, index code repositories — agents get RAG-powered access to search and reference your data.

Visibility is built in. Every agent action, tool call, and LLM request is tracked. See real-time activity, read agent reports, and monitor costs down to the per-request level.

Architecture at a glance

Tech stack

Layer	Technology
Frontend	Next.js 14, TypeScript, Tailwind CSS, shadcn/ui
Backend	Python, FastAPI, SQLAlchemy, PostgreSQL, Redis
AI	Multi-provider — OpenRouter, OpenAI, Anthropic, DeepSeek
Storage	AWS S3 for documents, S3 Vectors for embeddings
Auth	Clerk with full workspace isolation
Infra	Docker, Railway

Key concepts

Concept	What it is
Agent	An AI worker with a model, persona, tools, and skills
Workspace	An isolated environment for a team — agents, data, and config are scoped to it
Recipe	A multi-step workflow that coordinates multiple agents
Plugin	A marketplace-distributed package of skills, commands, and agent configs
Skill	A git-based capability loaded into an agent from a repository
Tool	An external app integration (GitHub, Slack, Jira) via Composio
Knowledge Base	A collection of documents indexed for semantic search (RAG)
Universal Router	The system that automatically sends messages to the right agent