What is Atreyu?
An environment for AI that can converse, act, and carry work forward.
A platform around the model
Atreyu is a privately developed AI orchestration platform. It brings conversations, model providers, agents, workflows, tools, memory, and media into a shared environment. The model generates a response; the platform manages the surrounding state, permissions, execution, and continuity.
From a conversation to real work
A conversation can invoke a tool backed by a versioned workflow. That workflow can coordinate model calls and authorized operations, record progress, and return an outcome to the conversation. Separate workers handle longer-running execution and context maintenance.
Who it is for
This public home is for people exploring the system, existing users finding their way in, and technically curious readers studying its architecture. Access to a running environment is separate from access to these documents.
Closed source, explained openly
Atreyu is currently closed source. These pages explain its design without distributing the implementation, credentials, private agent prompts, user conversations, or internal operational inventories. They are a public companion to the Architecture Atlas, not a platform source release. Our MI50-optimized llama.cpp inference engine is a separate, public MIT-licensed fork.
An evolving system
Features and model availability depend on the environment and account. This guide describes architectural boundaries; it is not an availability guarantee. Development and older semi-production deployments should not be assumed to have matching capabilities.