> For clean Markdown of any page, append .md to the page URL. > For a complete documentation index, see https://docs.fabricgateway.ai/llms.txt. > For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs.fabricgateway.ai/_mcp/server. # Understand key concepts and the console > Learn the terms that Fabric AI Gateway uses and find your way around the console. > **Info** > > This topic describes the product direction and implementation, but much of the platform is still evolving. Some details, capabilities, and decisions may change as development continues. Fabric AI Gateway manages how your applications, tools, and agents reach LLM providers, MCP servers, and internal, external, or public APIs. You connect a provider or service, expose it through a gateway route, and then attach plugins to the route to control authentication, routing, and other behavior. You can review every request in the console. This topic introduces the main concepts, shows you where to find things in the console, and outlines a typical setup path. ## Key concepts The console and the docs use the following terms: | Term | Meaning | | ---------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | **Organization** | The account you sign in to. You can belong to several organizations and switch between them. | | **Connection** | The link between the gateway and an upstream: an LLM provider, MCP server, REST API, or agent. It holds the target URL. Routes and credentials refer to a connection by its name. | | **Route** | The path your applications send requests to. A route exposes the endpoints you choose under a base path. | | **Gateway key** | A key that callers send with each request to authenticate to a route. | | **API key** | An administrative key for managing your Fabric AI Gateway organization programmatically. It is separate from a gateway key. For details, see [Manage API keys](/fabric/administration/api-keys). | | **Plugin** | A reusable behavior, such as authentication, rate limiting, or PII redaction, that you choose from the plugin catalog. When you configure a plugin, you apply it at a scope (globally, to a connection, to a route, or to a consumer or consumer group) and can add conditions that limit when it runs. Each configured copy is a plugin instance. | | **Vault** | A credential backend that the gateway uses to look up upstream credentials. | | **Namespace** | A space that separates teams' connections, services, analytics, traces, and users within an organization. For details, see [Manage namespaces](/fabric/administration/namespaces). | | **Data plane hosting** | Postman can host the data plane for you, or you can run it in your own cloud. For details, see [Deployment options](/get-started/architecture#deployment-options). | | **Data plane host** | The hostname that receives your applications' requests. The docs show it as `GATEWAY_HOST`. To get the URL for a route, select **Copy LLM base URL** on the **Routes** tab. | ## How the concepts fit together Each organization is isolated and contains connections. Each connection has one or more routes. You attach plugins to routes to control traffic. Some settings apply to your whole organization: members and their roles, invitations, and API keys. Admins manage these under **Admin**. Traces and analytics record the traffic that passes through your routes; you can view them for the current namespace or, under **Admin**, for the whole organization. ## Find your way around the console The sidebar has the following groups: | Group | Items | What you use them for | | --------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ | ----------------------------------------------------------------------------------------- | | **Gateway** | **LLMs**, **MCP servers**, **Services**, **Plugins** | Connect providers, review routes, and attach plugins. | | Namespace items | **Members**, **Credentials** | Manage the people and credentials in the current namespace. | | **Admin** | **Analytics**, **Traces**, **Namespaces**, **Members & Invites**, **API keys**, **Audit logs**, **Usage & Billing** (where billing is on), and, if you run your own data plane, **Gateway health**, **Single sign-on**, **Vaults**, and **Data plane** | Review organization-wide usage and requests, and manage access and organization settings. | The items you see depend on your permissions and on how your data plane is hosted. **Super admin** users see everything. **Member** users see the **Gateway** items, **Members**, and **Credentials**, but not **Admin**. **Gateway health**, **Single sign-on**, **Vaults**, and **Data plane** appear only if you run your own data plane. If an item described in these docs is missing, check your role and which organization you're in. Select your organization name at the bottom of the sidebar to switch. ## Set up your first route To go from an empty organization to a working, monitored route, complete the following tasks in order: 1. Connect a provider and create a route by following [Connect an LLM provider](/fabric/gateways/llm-providers). The wizard also creates a gateway key for the route. 2. Point your tools at the route by following [Use Claude Code with Fabric AI Gateway](/fabric/agents/claude-code). 3. Review the requests that pass through the route by following [Monitor gateway traffic](/fabric/observability). 4. Give your team access by following [Manage users and invitations](/fabric/administration/users). ## Next steps After your first route works, explore the following plugins in the plugin catalog: * **Adaptive Model Routing** picks the most cost-effective model for each LLM request. For the steps, see [Adaptive Model Routing and Provider Failover](/fabric/plugins/routing). * **Tool Search** lets agents search for the MCP tools they need instead of loading every tool from every server. * **LLM Rate Limit** enforces request and token limits. * **Killswitch** immediately blocks traffic for configured scopes.