What is Athena?

Athena is an AI agent orchestration platform. It gives your product a self-contained backend that hosts agents, manages users, handles credit billing, and serves CMS content — all behind a single API.

You build the app. Athena runs the AI.

Who Is Athena For?

Team	Use case
Mobile and web product teams	Add AI agents to an existing app without building LLM infrastructure
Platform teams	Host and govern multiple AI products from one workspace
Backend engineers	Server-to-server agent execution via workspace API keys

How It Works

Your app exchanges a user token for a short-lived API key, then calls an agent execution endpoint. Athena handles the LLM calls, workflow logic, credit billing, and memory — and streams results back as SSE events.

Your App
  │
  ├── POST /auth/external/{workspace_id}   ← exchange user JWT
  │       └── returns eak_ key (24 h)
  │
  ├── POST /agents/{id}/execute/stream     ← stream agent output (SSE)
  │       headers: X-API-Key: eak_...
  │
  └── GET  /my/credits                     ← check credit balance

Key Terms

Term	Meaning
Workspace	Tenant boundary — every agent, user, and credit ledger lives inside one
Agent	A published workflow with a system prompt, model, and tools
Execution	One run of an agent — holds message history, state, and credits
External Customer	Your end-user, identified by a Firebase UID or Clerk ID
`wak_`	Workspace API Key — long-lived, server-side only
`eak_`	Ephemeral API Key — short-lived, safe to pass to client devices

Next Steps

Concepts — workspace, agent, execution, and customer in depth
Architecture — system components and execution pipeline
Quickstart — zero to first agent execution in 5 minutes