Start Here
Platform Overview
Architecture, execution tiers, and developer foundation of the Omni Agent Platform.
Omni is an index-grounded, stateful AI agent platform that pairs global edge routing with an authoritative state engine to deliver low-latency completions, persistent semantic memory, and tool execution.
Architecture
Omni operates across unified infrastructure layers:
Initializing Diagram Engine...
Layer Breakdown
| Layer | Component | Function |
|---|---|---|
| 1. Clients | Web, Mobile, CLI | User-facing interfaces that render a standardized real-time SSE stream (thought, tool_call, tool_result, message, done). |
| 2. Edge Control Plane | edge.omnistatic.com | Globally distributed edge proxy handling API key verification, sliding-window rate limiting, memory RAG context assembly, and fast-path routing. |
| 3. Stateful Agent Engine | Authoritative Core | Executes multi-step LLM reasoning loops, autonomous tool invocations, and enforces the central margin-backed credit ledger. |
Status Taxonomy
Every capability in these docs is labeled with an explicit status indicator:
| Label | Meaning |
|---|---|
| Live | Production-proved and active. Available to all authenticated API keys. |
| Preview | Public API contract defined; undergoing phased developer rollout. |
| Coming Soon | Planned architecture and future compute tiers. |
Developer Principles
- Explicit Scopes: API keys receive only the permissions requested at creation (
chat,memory:read,memory:write,usage:read). - Canonical Developer Base URL: All developer integrations connect strictly to
https://edge.omnistatic.com. - Deterministic Memory Injection: Relevant memories and biographical notes are automatically retrieved and assembled into prompt context before model execution.
- Authoritative Quota Ledger: Usage counters are tracked with margin-backed limits, preventing unexpected token overruns.