ODOCK.AI
Getting Started

What is Odock?

Learn what Odock is, why teams use it, and where an AI governance gateway fits in LLM and MCP infrastructure.

What is Odock?

Odock is an AI governance gateway for LLM and MCP traffic.

It sits between your applications, agents, developer tools, and AI providers so every model call and tool call passes through one controlled entry point.

Instead of connecting each application directly to OpenAI, Anthropic, Google, Azure OpenAI, vLLM, custom models, or MCP servers, Odock centralizes how AI traffic is managed.

With Odock, teams can control:

  • who can access which models and MCP tools
  • which provider keys are used
  • how requests are routed
  • how budgets, quotas, and rate limits are enforced
  • how usage, cost, latency, and errors are tracked
  • which security checks run before and after requests

Why Use Odock?

AI adoption often starts with a few direct API calls. That works for experiments, but it becomes hard to manage when multiple teams, providers, models, agents, and MCP tools move into production.

Without a gateway, API keys are copied across services, costs are hard to attribute, model access is difficult to control, and security policies depend on each application.

Odock solves this by putting governance directly in the request path. Every call is authenticated, checked, routed, recorded, and monitored before it reaches an upstream provider or MCP server.

Where Odock Fits

Odock is useful when AI traffic becomes shared infrastructure.

Centralize LLM Access

Expose approved models to all teams through one governed gateway instead of scattered direct provider calls.

Protect Provider Keys

Keep OpenAI, Anthropic, Google, Azure OpenAI, vLLM, and custom provider credentials behind Odock.

Give Developers One Entry Point

Support OpenAI-compatible, Anthropic-compatible, Gemini-compatible, vLLM-compatible, and unified Odock endpoints.

Control Access

Define who can use each model or MCP tool by organisation, team, user, or virtual API key.

Manage Spend

Use budgets, quotas, reservations, and usage records to control consumption before costs grow.

Govern MCP Tools

Control which agents and API keys can access MCP servers, tools, and semantic rules.

Route Without Rewrites

Change providers, models, or routing policies without redeploying every application.

Add Security and Visibility

Apply guardrails, logs, metrics, traces, and audit records around every AI request.

What Odock Controls

AreaPurpose
Identity and accessManage organisations, users, teams, roles, virtual API keys, model grants, and MCP grants
Provider managementStore provider configuration and encrypted provider keys centrally
Cost controlsEnforce budgets, quotas, reservations, and usage tracking
RoutingDecide which provider or model handles each request
SecurityApply policies, rate limits, SafetySec checks, and plugin hooks
ObservabilityTrack logs, metrics, latency, errors, routing attempts, and usage

On this page