Skip to main content

​
What is AMAS?

AMAS — the Artaios Multi Agent System — is a declarative system for building multi-agent AI systems. You define an agentic system in YAML: agents, models, prompts, tools, knowledge, memory and the orchestration between them. AMAS runs it, and exposes it over REST and WebSocket. You work with AMAS two ways, interchangeably:
  • Agent Studio — AMAS’s web interface. Write YAML, run it in the Playground, build knowledge bases, manage keys.
  • The API — REST and WebSocket, taking the same YAML.
AMAS, Agent Studio and Artaios are three different things. AMAS is the system and the YAML. Agent Studio is AMAS’s UI. Artaios is the separate end-user chat product that finished agents can be published to — it is a destination, not a build surface. Full breakdown: Concepts & Glossary.

​
Who is this for?

These docs are written for hands-on technical builders — people who would rather write a config than click through a wizard, and who want to see and control every part of an agentic system.
  • Developers building AI-powered applications
  • AI engineers and consultants deploying custom agent solutions for clients
  • Technical architects prototyping multi-agent workflows
This is not a no-code product. There is no drag-and-drop canvas. Every agentic system in AMAS is a YAML file you write and maintain yourself.We assume you are comfortable with:
  • YAML — nesting, lists, block scalars, anchors
  • HTTP APIs — authentication headers, JSON request and response bodies
  • Prompt engineering — you will be writing and iterating on system prompts
  • Reading reference docs — these pages are a reference, not a course
If you want point-and-click agent building for non-technical users, AMAS is the wrong layer. The Artaios chat product is where non-technical colleagues consume what you build here.

​
Core Capabilities

Rapid Development

Define an agentic system in YAML and run it immediately

Multi-Agent Orchestration

Supervisor, sequential and conditional patterns out of the box

Staying Ahead of the AI Wave

New models and multi-agent techniques land as config options, not rewrites

Bring your Data to Life

Upload documents and connect data sources from your existing systems

Integrate with External Services

Toolkits, MCP servers and your own custom tools

Production Monitoring

Built-in tracing and observability on every run

​
Architecture

AMAS runs on German cloud infrastructure (Open Telekom Cloud) operated by arttac solutions, ensuring data sovereignty and compliance with European regulations. Your agentic systems can be:
  • Tested in the Agent Studio Playground with instant feedback
  • Deployed via REST API for server-side integration
  • Streamed via WebSocket for real-time applications
  • Embedded as public chatbots on your websites
  • Published to the Artaios chat product for colleagues to use directly

​
Getting Access

AMAS is currently in closed beta.

Request Access

Once approved, you’ll receive credentials for Agent Studio at amas.artaios.com.

​
Your Feedback Matters

As an early user, your feedback directly shapes the system. Throughout these docs, you’ll see prompts to share your thoughts — please do! Your input helps us build something genuinely useful.

​
Next Steps

Concepts & Glossary

AMAS vs Agent Studio vs Artaios, and the vocabulary

Quickstart Guide

Build and test your first agent in 10 minutes

YAML Configuration

Learn how to configure agents with YAML

Connecting Your Agent

REST, WebSocket and embedded chatbot

​
What Makes This Different?

Unlike point-and-click AI builders or code-heavy frameworks, AMAS sits in the middle:
  • Configuration over Code: YAML-based setup means readable, version-controllable configs
  • Built for Multi-Agent: Design complex workflows with agent coordination out of the box
  • Observability First: See what your agents are doing with comprehensive tracing
  • Integration Ready: Deploy via API, WebSocket, or embedded chatbot in minutes
If you’re building production AI systems and want control without complexity, you’re in the right place.