diff --git a/docs/ai/install-client-sdk.md b/docs/ai/install-client-sdk.md
index 120d88ec..0301d8d7 100644
--- a/docs/ai/install-client-sdk.md
+++ b/docs/ai/install-client-sdk.md
@@ -45,6 +45,8 @@ The AI Client SDK is written in TypeScript and includes complete type definition
 
 ## Framework-Specific Setup
 
+For `RevealSdkClient.initialize()` authentication and request header options, see [Request Configuration](/ai/sdk-request-configuration).
+
 ### Vanilla JavaScript
 
 #### Using ES Modules
diff --git a/docs/ai/sdk-overview.md b/docs/ai/sdk-overview.md
index 46396ae7..2d24d2ea 100644
--- a/docs/ai/sdk-overview.md
+++ b/docs/ai/sdk-overview.md
@@ -33,6 +33,8 @@ The client SDK requires the [AI Server SDK](/ai/install-server-sdk) to be instal
 
 :::
 
+If your AI server requires authentication or additional request headers, see [Request Configuration](/ai/sdk-request-configuration).
+
 ## Getting the Client Instance
 
 Once initialized, get the shared client instance anywhere in your application:
diff --git a/docs/ai/sdk-request-configuration.md b/docs/ai/sdk-request-configuration.md
new file mode 100644
index 00000000..a5df38ef
--- /dev/null
+++ b/docs/ai/sdk-request-configuration.md
@@ -0,0 +1,72 @@
+---
+sidebar_label: Request Configuration
+sidebar_custom_props:
+  status: preview
+---
+
+# Request Configuration
+
+:::info Preview
+
+Request configuration is a preview feature. It works as described below, but the option names and shapes may still change.
+
+:::
+
+`RevealSdkClient.initialize()` accepts options that control the HTTP requests the client sends to your AI server. Configure them only when your application needs authentication or additional request headers.
+
+## Bearer-Token Authentication
+
+Use `bearerToken` when the token is already available at startup. The client sends it as an `Authorization: Bearer <token>` header on every request.
+
+```typescript
+RevealSdkClient.initialize({
+  hostUrl: 'https://your-server.com',
+  bearerToken: 'your-token'
+});
+```
+
+To replace that token later, for example after a refresh, call `setBearerToken()` on the client instance:
+
+```typescript
+RevealSdkClient.getInstance().setBearerToken(newToken);
+```
+
+Use `getBearerToken` instead when the token must be resolved per request. The callback can be synchronous or asynchronous, and it is called before each request is sent, so your authentication provider can return or refresh the current token.
+
+```typescript
+RevealSdkClient.initialize({
+  hostUrl: 'https://your-server.com',
+  getBearerToken: async () => {
+    return await authService.getValidAccessToken();
+  }
+});
+```
+
+Use either the static token or the callback, not both.
+
+## Custom Request Headers
+
+Use `onRequest` when you need to add or modify headers other than the bearer token. The interceptor receives the outgoing request, can be asynchronous, and is called before each request is sent.
+
+```typescript
+RevealSdkClient.initialize({
+  hostUrl: 'https://your-server.com',
+  onRequest: async (request) => ({
+    ...request,
+    headers: {
+      ...request.headers,
+      'X-Tenant-Id': await tenantService.getCurrentTenantId()
+    }
+  })
+});
+```
+
+## Streaming Requests
+
+Bearer tokens and `onRequest` also apply to [streaming](/ai/sdk-streaming) requests. They are resolved once, when the Server-Sent Events (SSE) connection is opened, and not for each streamed event.
+
+## Response and Error Interceptors
+
+`initialize()` also accepts an `onResponse` interceptor, called with the raw `Response` of each completed request, and an `onError` interceptor, called with the error of each failed request. Both can be asynchronous, and both return the value the client continues with, so `onError` can enrich an error but cannot suppress it.
+
+For handling AI errors in application code, see [Error Handling](/ai/sdk-error-handling).
diff --git a/sidebars.ts b/sidebars.ts
index 77af8cf6..8dd4d545 100644
--- a/sidebars.ts
+++ b/sidebars.ts
@@ -296,6 +296,7 @@ const sidebars: SidebarsConfig = {
         { type: "doc", label: "Chat", id: "ai/sdk-chat" },
         { type: "doc", label: "Streaming Responses", id: "ai/sdk-streaming" },
         { type: "doc", label: "Error Handling", id: "ai/sdk-error-handling" },
+        { type: "doc", label: "Request Configuration", id: "ai/sdk-request-configuration" },
       ]
     },