diff --git a/app/schema.json b/app/schema.json
index 93e5bf06a..2bed0647b 100644
--- a/app/schema.json
+++ b/app/schema.json
@@ -1218,332 +1218,6 @@
         }
       ]
     },
-    {
-      "description": "Run one-shot agent chat with optional model overrides.",
-      "function": "agent_chat",
-      "inputs": [
-        { "comment": "User message.", "name": "message", "required": true, "ty": "String" },
-        {
-          "comment": "Optional model override.",
-          "name": "model_override",
-          "required": false,
-          "ty": { "Option": "String" }
-        },
-        {
-          "comment": "Optional temperature override.",
-          "name": "temperature",
-          "required": false,
-          "ty": { "Option": "F64" }
-        }
-      ],
-      "method": "openhuman.local_ai_agent_chat",
-      "namespace": "local_ai",
-      "outputs": [
-        { "comment": "Agent response payload.", "name": "response", "required": true, "ty": "Json" }
-      ]
-    },
-    {
-      "description": "Run one-shot lightweight provider chat.",
-      "function": "agent_chat_simple",
-      "inputs": [
-        { "comment": "User message.", "name": "message", "required": true, "ty": "String" },
-        {
-          "comment": "Optional model override.",
-          "name": "model_override",
-          "required": false,
-          "ty": { "Option": "String" }
-        },
-        {
-          "comment": "Optional temperature override.",
-          "name": "temperature",
-          "required": false,
-          "ty": { "Option": "F64" }
-        }
-      ],
-      "method": "openhuman.local_ai_agent_chat_simple",
-      "namespace": "local_ai",
-      "outputs": [
-        { "comment": "Agent response payload.", "name": "response", "required": true, "ty": "Json" }
-      ]
-    },
-    {
-      "description": "Terminate REPL session.",
-      "function": "agent_repl_session_end",
-      "inputs": [
-        { "comment": "REPL session id.", "name": "session_id", "required": true, "ty": "String" }
-      ],
-      "method": "openhuman.local_ai_agent_repl_session_end",
-      "namespace": "local_ai",
-      "outputs": [
-        { "comment": "Session end result.", "name": "result", "required": true, "ty": "Json" }
-      ]
-    },
-    {
-      "description": "Clear REPL session history.",
-      "function": "agent_repl_session_reset",
-      "inputs": [
-        { "comment": "REPL session id.", "name": "session_id", "required": true, "ty": "String" }
-      ],
-      "method": "openhuman.local_ai_agent_repl_session_reset",
-      "namespace": "local_ai",
-      "outputs": [
-        { "comment": "Session reset result.", "name": "result", "required": true, "ty": "Json" }
-      ]
-    },
-    {
-      "description": "Create a persistent REPL agent session.",
-      "function": "agent_repl_session_start",
-      "inputs": [
-        {
-          "comment": "Optional session id.",
-          "name": "session_id",
-          "required": false,
-          "ty": { "Option": "String" }
-        },
-        {
-          "comment": "Optional model override.",
-          "name": "model_override",
-          "required": false,
-          "ty": { "Option": "String" }
-        },
-        {
-          "comment": "Optional temperature override.",
-          "name": "temperature",
-          "required": false,
-          "ty": { "Option": "F64" }
-        }
-      ],
-      "method": "openhuman.local_ai_agent_repl_session_start",
-      "namespace": "local_ai",
-      "outputs": [
-        { "comment": "Session creation result.", "name": "result", "required": true, "ty": "Json" }
-      ]
-    },
-    {
-      "description": "Get local AI asset installation status.",
-      "function": "assets_status",
-      "inputs": [],
-      "method": "openhuman.local_ai_assets_status",
-      "namespace": "local_ai",
-      "outputs": [
-        { "comment": "Assets status payload.", "name": "status", "required": true, "ty": "Json" }
-      ]
-    },
-    {
-      "description": "Trigger local AI model download bootstrap.",
-      "function": "download",
-      "inputs": [
-        {
-          "comment": "Reset state before download.",
-          "name": "force",
-          "required": false,
-          "ty": { "Option": "Bool" }
-        }
-      ],
-      "method": "openhuman.local_ai_download",
-      "namespace": "local_ai",
-      "outputs": [
-        { "comment": "Local AI status payload.", "name": "status", "required": true, "ty": "Json" }
-      ]
-    },
-    {
-      "description": "Trigger full local AI asset download.",
-      "function": "download_all_assets",
-      "inputs": [
-        {
-          "comment": "Reset state before download.",
-          "name": "force",
-          "required": false,
-          "ty": { "Option": "Bool" }
-        }
-      ],
-      "method": "openhuman.local_ai_download_all_assets",
-      "namespace": "local_ai",
-      "outputs": [
-        {
-          "comment": "Download progress payload.",
-          "name": "progress",
-          "required": true,
-          "ty": "Json"
-        }
-      ]
-    },
-    {
-      "description": "Trigger download for one local AI asset capability.",
-      "function": "download_asset",
-      "inputs": [
-        {
-          "comment": "Asset capability id.",
-          "name": "capability",
-          "required": true,
-          "ty": "String"
-        }
-      ],
-      "method": "openhuman.local_ai_download_asset",
-      "namespace": "local_ai",
-      "outputs": [
-        { "comment": "Assets status payload.", "name": "status", "required": true, "ty": "Json" }
-      ]
-    },
-    {
-      "description": "Get local AI download progress.",
-      "function": "downloads_progress",
-      "inputs": [],
-      "method": "openhuman.local_ai_downloads_progress",
-      "namespace": "local_ai",
-      "outputs": [
-        {
-          "comment": "Download progress payload.",
-          "name": "progress",
-          "required": true,
-          "ty": "Json"
-        }
-      ]
-    },
-    {
-      "description": "Generate embeddings for text inputs.",
-      "function": "embed",
-      "inputs": [
-        {
-          "comment": "Texts to embed.",
-          "name": "inputs",
-          "required": true,
-          "ty": { "Array": "String" }
-        }
-      ],
-      "method": "openhuman.local_ai_embed",
-      "namespace": "local_ai",
-      "outputs": [
-        {
-          "comment": "Embedding result payload.",
-          "name": "embedding",
-          "required": true,
-          "ty": "Json"
-        }
-      ]
-    },
-    {
-      "description": "Run direct local AI prompt.",
-      "function": "prompt",
-      "inputs": [
-        { "comment": "Prompt text.", "name": "prompt", "required": true, "ty": "String" },
-        {
-          "comment": "Optional max output tokens.",
-          "name": "max_tokens",
-          "required": false,
-          "ty": { "Option": "U64" }
-        },
-        {
-          "comment": "Disable thinking mode.",
-          "name": "no_think",
-          "required": false,
-          "ty": { "Option": "Bool" }
-        }
-      ],
-      "method": "openhuman.local_ai_prompt",
-      "namespace": "local_ai",
-      "outputs": [
-        { "comment": "Prompt output text.", "name": "output", "required": true, "ty": "Json" }
-      ]
-    },
-    {
-      "description": "Read local AI service status.",
-      "function": "status",
-      "inputs": [],
-      "method": "openhuman.local_ai_status",
-      "namespace": "local_ai",
-      "outputs": [
-        { "comment": "Local AI status payload.", "name": "status", "required": true, "ty": "Json" }
-      ]
-    },
-    {
-      "description": "Summarize text with local AI model.",
-      "function": "summarize",
-      "inputs": [
-        { "comment": "Input text.", "name": "text", "required": true, "ty": "String" },
-        {
-          "comment": "Optional max output tokens.",
-          "name": "max_tokens",
-          "required": false,
-          "ty": { "Option": "U64" }
-        }
-      ],
-      "method": "openhuman.local_ai_summarize",
-      "namespace": "local_ai",
-      "outputs": [{ "comment": "Summary text.", "name": "summary", "required": true, "ty": "Json" }]
-    },
-    {
-      "description": "Transcribe audio from file path.",
-      "function": "transcribe",
-      "inputs": [
-        { "comment": "Input audio path.", "name": "audio_path", "required": true, "ty": "String" }
-      ],
-      "method": "openhuman.local_ai_transcribe",
-      "namespace": "local_ai",
-      "outputs": [
-        { "comment": "Transcription payload.", "name": "speech", "required": true, "ty": "Json" }
-      ]
-    },
-    {
-      "description": "Transcribe audio from raw bytes.",
-      "function": "transcribe_bytes",
-      "inputs": [
-        { "comment": "Raw audio bytes.", "name": "audio_bytes", "required": true, "ty": "Bytes" },
-        {
-          "comment": "Optional audio extension.",
-          "name": "extension",
-          "required": false,
-          "ty": { "Option": "String" }
-        }
-      ],
-      "method": "openhuman.local_ai_transcribe_bytes",
-      "namespace": "local_ai",
-      "outputs": [
-        { "comment": "Transcription payload.", "name": "speech", "required": true, "ty": "Json" }
-      ]
-    },
-    {
-      "description": "Synthesize speech from text.",
-      "function": "tts",
-      "inputs": [
-        { "comment": "Input text.", "name": "text", "required": true, "ty": "String" },
-        {
-          "comment": "Optional output path.",
-          "name": "output_path",
-          "required": false,
-          "ty": { "Option": "String" }
-        }
-      ],
-      "method": "openhuman.local_ai_tts",
-      "namespace": "local_ai",
-      "outputs": [
-        { "comment": "TTS result payload.", "name": "tts", "required": true, "ty": "Json" }
-      ]
-    },
-    {
-      "description": "Run multimodal local AI prompt with image refs.",
-      "function": "vision_prompt",
-      "inputs": [
-        { "comment": "Prompt text.", "name": "prompt", "required": true, "ty": "String" },
-        {
-          "comment": "Image references to include.",
-          "name": "image_refs",
-          "required": true,
-          "ty": { "Array": "String" }
-        },
-        {
-          "comment": "Optional max output tokens.",
-          "name": "max_tokens",
-          "required": false,
-          "ty": { "Option": "U64" }
-        }
-      ],
-      "method": "openhuman.local_ai_vision_prompt",
-      "namespace": "local_ai",
-      "outputs": [
-        { "comment": "Prompt output text.", "name": "output", "required": true, "ty": "Json" }
-      ]
-    },
     {
       "description": "Migrate OpenClaw memory into current workspace.",
       "function": "openclaw",
diff --git a/app/src/services/__tests__/rpcMethods.test.ts b/app/src/services/__tests__/rpcMethods.test.ts
index 7512ffb14..10ee46b39 100644
--- a/app/src/services/__tests__/rpcMethods.test.ts
+++ b/app/src/services/__tests__/rpcMethods.test.ts
@@ -131,6 +131,10 @@ describe('rpcMethods catalog', () => {
         path.resolve(__dirname, '../../../../src/openhuman/inference/schemas.rs'),
         'utf8'
       ),
+      fs.readFileSync(
+        path.resolve(__dirname, '../../../../src/openhuman/inference/local/schemas.rs'),
+        'utf8'
+      ),
       fs.readFileSync(
         path.resolve(__dirname, '../../../../src/openhuman/embeddings/schemas.rs'),
         'utf8'
diff --git a/app/src/services/api/__tests__/voiceInstallApi.test.ts b/app/src/services/api/__tests__/voiceInstallApi.test.ts
index ce7f51c02..3e3cb4078 100644
--- a/app/src/services/api/__tests__/voiceInstallApi.test.ts
+++ b/app/src/services/api/__tests__/voiceInstallApi.test.ts
@@ -33,7 +33,7 @@ describe('voiceInstallApi', () => {
       vi.mocked(callCoreRpc).mockResolvedValueOnce(buildStatus({ engine: 'whisper' }));
       const result = await installWhisper({ modelSize: 'tiny', force: true });
       expect(callCoreRpc).toHaveBeenCalledWith({
-        method: 'openhuman.local_ai_install_whisper',
+        method: 'openhuman.inference_install_whisper',
         params: { model_size: 'tiny', force: true },
       });
       expect(result.engine).toBe('whisper');
@@ -45,7 +45,7 @@ describe('voiceInstallApi', () => {
       vi.mocked(callCoreRpc).mockResolvedValueOnce(buildStatus());
       await installWhisper();
       expect(callCoreRpc).toHaveBeenCalledWith({
-        method: 'openhuman.local_ai_install_whisper',
+        method: 'openhuman.inference_install_whisper',
         params: { model_size: undefined, force: undefined },
       });
     });
@@ -65,7 +65,7 @@ describe('voiceInstallApi', () => {
       );
       const result = await installPiper({ voiceId: 'en_US-lessac-medium', force: false });
       expect(callCoreRpc).toHaveBeenCalledWith({
-        method: 'openhuman.local_ai_install_piper',
+        method: 'openhuman.inference_install_piper',
         params: { voice_id: 'en_US-lessac-medium', force: false },
       });
       expect(result.state).toBe('installing');
@@ -77,7 +77,7 @@ describe('voiceInstallApi', () => {
       vi.mocked(callCoreRpc).mockResolvedValueOnce(buildStatus({ engine: 'piper' }));
       await installPiper();
       expect(callCoreRpc).toHaveBeenCalledWith({
-        method: 'openhuman.local_ai_install_piper',
+        method: 'openhuman.inference_install_piper',
         params: { voice_id: undefined, force: undefined },
       });
     });
@@ -91,7 +91,7 @@ describe('voiceInstallApi', () => {
       );
       const result = await whisperInstallStatus();
       expect(callCoreRpc).toHaveBeenCalledWith({
-        method: 'openhuman.local_ai_whisper_install_status',
+        method: 'openhuman.inference_whisper_install_status',
         params: {},
       });
       expect(result.state).toBe('missing');
@@ -106,7 +106,7 @@ describe('voiceInstallApi', () => {
       );
       const result = await piperInstallStatus();
       expect(callCoreRpc).toHaveBeenCalledWith({
-        method: 'openhuman.local_ai_piper_install_status',
+        method: 'openhuman.inference_piper_install_status',
         params: {},
       });
       expect(result.state).toBe('error');
diff --git a/app/src/services/api/voiceInstallApi.ts b/app/src/services/api/voiceInstallApi.ts
index 13c8de4d1..14fadaf40 100644
--- a/app/src/services/api/voiceInstallApi.ts
+++ b/app/src/services/api/voiceInstallApi.ts
@@ -1,5 +1,5 @@
 /**
- * Voice engine installer API — wraps the four new `local_ai.*` RPCs that
+ * Voice engine installer API — wraps the `inference.*` RPCs that
  * orchestrate downloads of the Whisper GGML model + binary and the Piper
  * binary + bundled voice into the workspace.
  *
@@ -15,7 +15,7 @@ import { callCoreRpc } from '../coreRpcClient';
 const log = debug('voiceInstallApi');
 
 /**
- * Stable wire shape of [`crate::openhuman::local_ai::voice_install_common::VoiceInstallState`].
+ * Stable wire shape of [`crate::openhuman::inference::local::voice_install_common::VoiceInstallState`].
  *
  * The Rust enum serializes via `#[serde(rename_all = "snake_case")]` so
  * the TypeScript union mirrors the lowercase variants exactly.
@@ -67,7 +67,7 @@ export async function installWhisper(
 ): Promise<VoiceInstallStatus> {
   log('[voice-install:whisper] kick-off %o', params);
   const result = await callCoreRpc<VoiceInstallStatus>({
-    method: 'openhuman.local_ai_install_whisper',
+    method: 'openhuman.inference_install_whisper',
     params: { model_size: params.modelSize, force: params.force },
   });
   log('[voice-install:whisper] result state=%s stage=%s', result.state, result.stage ?? '<none>');
@@ -81,7 +81,7 @@ export async function installWhisper(
 export async function installPiper(params: InstallPiperParams = {}): Promise<VoiceInstallStatus> {
   log('[voice-install:piper] kick-off %o', params);
   const result = await callCoreRpc<VoiceInstallStatus>({
-    method: 'openhuman.local_ai_install_piper',
+    method: 'openhuman.inference_install_piper',
     params: { voice_id: params.voiceId, force: params.force },
   });
   log('[voice-install:piper] result state=%s stage=%s', result.state, result.stage ?? '<none>');
@@ -96,7 +96,7 @@ export async function installPiper(params: InstallPiperParams = {}): Promise<Voi
  */
 export async function whisperInstallStatus(): Promise<VoiceInstallStatus> {
   return await callCoreRpc<VoiceInstallStatus>({
-    method: 'openhuman.local_ai_whisper_install_status',
+    method: 'openhuman.inference_whisper_install_status',
     params: {},
   });
 }
@@ -107,7 +107,7 @@ export async function whisperInstallStatus(): Promise<VoiceInstallStatus> {
  */
 export async function piperInstallStatus(): Promise<VoiceInstallStatus> {
   return await callCoreRpc<VoiceInstallStatus>({
-    method: 'openhuman.local_ai_piper_install_status',
+    method: 'openhuman.inference_piper_install_status',
     params: {},
   });
 }
diff --git a/app/src/services/rpcMethods.ts b/app/src/services/rpcMethods.ts
index e7417a9fe..7be9a3d93 100644
--- a/app/src/services/rpcMethods.ts
+++ b/app/src/services/rpcMethods.ts
@@ -28,14 +28,27 @@ export const CORE_RPC_METHODS = {
   configWorkspaceOnboardingFlagExists: 'openhuman.config_workspace_onboarding_flag_exists',
   configWorkspaceOnboardingFlagSet: 'openhuman.config_workspace_onboarding_flag_set',
   corePing: 'core.ping',
+  inferenceAgentChat: 'openhuman.inference_agent_chat',
+  inferenceAgentChatSimple: 'openhuman.inference_agent_chat_simple',
   inferenceApplyPreset: 'openhuman.inference_apply_preset',
+  inferenceAssetsStatus: 'openhuman.inference_assets_status',
   inferenceDiagnostics: 'openhuman.inference_diagnostics',
   inferenceDeviceProfile: 'openhuman.inference_device_profile',
+  inferenceDownloadAsset: 'openhuman.inference_download_asset',
+  inferenceDownloadsProgress: 'openhuman.inference_downloads_progress',
   inferenceGetClientConfig: 'openhuman.inference_get_client_config',
+  inferenceInstallPiper: 'openhuman.inference_install_piper',
+  inferenceInstallWhisper: 'openhuman.inference_install_whisper',
   inferenceListModels: 'openhuman.inference_list_models',
+  inferencePiperInstallStatus: 'openhuman.inference_piper_install_status',
   inferencePresets: 'openhuman.inference_presets',
+  inferenceTestConnection: 'openhuman.inference_test_connection',
+  inferenceTranscribe: 'openhuman.inference_transcribe',
+  inferenceTranscribeBytes: 'openhuman.inference_transcribe_bytes',
+  inferenceTts: 'openhuman.inference_tts',
   inferenceUpdateLocalSettings: 'openhuman.inference_update_local_settings',
   inferenceUpdateModelSettings: 'openhuman.inference_update_model_settings',
+  inferenceWhisperInstallStatus: 'openhuman.inference_whisper_install_status',
   providersListModels: 'openhuman.inference_list_models',
   screenIntelligenceStatus: 'openhuman.screen_intelligence_status',
   embeddingsGetSettings: 'openhuman.embeddings_get_settings',
@@ -81,10 +94,23 @@ export const LEGACY_METHOD_ALIASES: Record<string, CoreRpcMethod> = {
   'openhuman.workspace_onboarding_flag_exists':
     CORE_RPC_METHODS.configWorkspaceOnboardingFlagExists,
   'openhuman.workspace_onboarding_flag_set': CORE_RPC_METHODS.configWorkspaceOnboardingFlagSet,
+  'openhuman.local_ai_agent_chat': CORE_RPC_METHODS.inferenceAgentChat,
+  'openhuman.local_ai_agent_chat_simple': CORE_RPC_METHODS.inferenceAgentChatSimple,
   'openhuman.local_ai_apply_preset': CORE_RPC_METHODS.inferenceApplyPreset,
+  'openhuman.local_ai_assets_status': CORE_RPC_METHODS.inferenceAssetsStatus,
   'openhuman.local_ai_device_profile': CORE_RPC_METHODS.inferenceDeviceProfile,
   'openhuman.local_ai_diagnostics': CORE_RPC_METHODS.inferenceDiagnostics,
+  'openhuman.local_ai_download_asset': CORE_RPC_METHODS.inferenceDownloadAsset,
+  'openhuman.local_ai_downloads_progress': CORE_RPC_METHODS.inferenceDownloadsProgress,
+  'openhuman.local_ai_install_piper': CORE_RPC_METHODS.inferenceInstallPiper,
+  'openhuman.local_ai_install_whisper': CORE_RPC_METHODS.inferenceInstallWhisper,
+  'openhuman.local_ai_piper_install_status': CORE_RPC_METHODS.inferencePiperInstallStatus,
   'openhuman.local_ai_presets': CORE_RPC_METHODS.inferencePresets,
+  'openhuman.local_ai_test_connection': CORE_RPC_METHODS.inferenceTestConnection,
+  'openhuman.local_ai_transcribe': CORE_RPC_METHODS.inferenceTranscribe,
+  'openhuman.local_ai_transcribe_bytes': CORE_RPC_METHODS.inferenceTranscribeBytes,
+  'openhuman.local_ai_tts': CORE_RPC_METHODS.inferenceTts,
+  'openhuman.local_ai_whisper_install_status': CORE_RPC_METHODS.inferenceWhisperInstallStatus,
   'openhuman.providers_list_models': CORE_RPC_METHODS.inferenceListModels,
   'openhuman.inference_embed': CORE_RPC_METHODS.embeddingsEmbed,
   health_snapshot: CORE_RPC_METHODS.healthSnapshot,
diff --git a/app/src/utils/tauriCommands/localAi.test.ts b/app/src/utils/tauriCommands/localAi.test.ts
index dccb8816c..fb5d88c8d 100644
--- a/app/src/utils/tauriCommands/localAi.test.ts
+++ b/app/src/utils/tauriCommands/localAi.test.ts
@@ -7,7 +7,7 @@ describe('openhumanLocalAiTestConnection', () => {
     vi.clearAllMocks();
   });
 
-  it('calls callCoreRpc with local_ai_test_connection method and url param', async () => {
+  it('calls callCoreRpc with inference_test_connection method and url param', async () => {
     const { callCoreRpc } = await import('../../services/coreRpcClient');
     const mockCallCoreRpc = callCoreRpc as ReturnType<typeof vi.fn>;
     mockCallCoreRpc.mockResolvedValueOnce({ reachable: true, models_count: 4 });
@@ -16,7 +16,7 @@ describe('openhumanLocalAiTestConnection', () => {
     const result = await openhumanLocalAiTestConnection('http://localhost:11434');
 
     expect(mockCallCoreRpc).toHaveBeenCalledWith({
-      method: 'openhuman.local_ai_test_connection',
+      method: 'openhuman.inference_test_connection',
       params: { url: 'http://localhost:11434' },
     });
     expect(result).toEqual({ reachable: true, models_count: 4 });
diff --git a/app/src/utils/tauriCommands/localAi.ts b/app/src/utils/tauriCommands/localAi.ts
index c7cccb3e5..e58711b32 100644
--- a/app/src/utils/tauriCommands/localAi.ts
+++ b/app/src/utils/tauriCommands/localAi.ts
@@ -298,7 +298,7 @@ export async function openhumanLocalAiTranscribe(
   audioPath: string
 ): Promise<CommandResponse<LocalAiSpeechResult>> {
   return await callCoreRpc<CommandResponse<LocalAiSpeechResult>>({
-    method: 'openhuman.local_ai_transcribe',
+    method: 'openhuman.inference_transcribe',
     params: { audio_path: audioPath },
   });
 }
@@ -308,7 +308,7 @@ export async function openhumanLocalAiTranscribeBytes(
   extension?: string
 ): Promise<CommandResponse<LocalAiSpeechResult>> {
   return await callCoreRpc<CommandResponse<LocalAiSpeechResult>>({
-    method: 'openhuman.local_ai_transcribe_bytes',
+    method: 'openhuman.inference_transcribe_bytes',
     params: { audio_bytes: audioBytes, extension },
   });
 }
@@ -318,7 +318,7 @@ export async function openhumanLocalAiTts(
   outputPath?: string
 ): Promise<CommandResponse<LocalAiTtsResult>> {
   return await callCoreRpc<CommandResponse<LocalAiTtsResult>>({
-    method: 'openhuman.local_ai_tts',
+    method: 'openhuman.inference_tts',
     params: { text, output_path: outputPath },
   });
 }
@@ -354,7 +354,7 @@ export async function openhumanLocalAiAssetsStatus(): Promise<
   CommandResponse<LocalAiAssetsStatus>
 > {
   return await callCoreRpc<CommandResponse<LocalAiAssetsStatus>>({
-    method: 'openhuman.local_ai_assets_status',
+    method: 'openhuman.inference_assets_status',
   });
 }
 
@@ -362,7 +362,7 @@ export async function openhumanLocalAiDownloadsProgress(): Promise<
   CommandResponse<LocalAiDownloadsProgress>
 > {
   return await callCoreRpc<CommandResponse<LocalAiDownloadsProgress>>({
-    method: 'openhuman.local_ai_downloads_progress',
+    method: 'openhuman.inference_downloads_progress',
   });
 }
 
@@ -370,7 +370,7 @@ export async function openhumanLocalAiDownloadAsset(
   capability: 'chat' | 'vision' | 'embedding' | 'stt' | 'tts'
 ): Promise<CommandResponse<LocalAiAssetsStatus>> {
   return await callCoreRpc<CommandResponse<LocalAiAssetsStatus>>({
-    method: 'openhuman.local_ai_download_asset',
+    method: 'openhuman.inference_download_asset',
     params: { capability },
   });
 }
@@ -407,7 +407,7 @@ export async function openhumanLocalAiTestConnection(
   url: string
 ): Promise<OllamaConnectionTestResult> {
   return await callCoreRpc<OllamaConnectionTestResult>({
-    method: 'openhuman.local_ai_test_connection',
+    method: 'openhuman.inference_test_connection',
     params: { url },
   });
 }
diff --git a/gitbooks/features/model-routing/local-ai.md b/gitbooks/features/model-routing/local-ai.md
index 643a585a9..59c95122b 100644
--- a/gitbooks/features/model-routing/local-ai.md
+++ b/gitbooks/features/model-routing/local-ai.md
@@ -90,7 +90,7 @@ The `local_ai.usage.*` booleans are consulted only during preset application and
 
 Prefer setting the `*_provider` fields directly when editing configuration by hand.
 
-In the desktop app, **Settings → AI & Skills → Local AI** exposes presets, pick one ("embeddings only", "memory + reflection", "everything local") and the right combination of flags is set for you. Status (Ollama reachability, model availability, per-subsystem enablement) is surfaced live via `openhuman.local_ai_status`.
+In the desktop app, **Settings → AI & Skills → Local AI** exposes presets, pick one ("embeddings only", "memory + reflection", "everything local") and the right combination of flags is set for you. Status (Ollama reachability, model availability, per-subsystem enablement) is surfaced live via `openhuman.inference_status`.
 
 ## When to turn it on
 
@@ -108,7 +108,7 @@ It is **not** worth turning on if you only have a few sources connected, the clo
 - Enough disk for the models (`gemma3:1b-it-qat` \~700 MB, `all-minilm:latest` \~23 MB).
 - Enough RAM to keep the model resident (8 GB+ recommended, 16 GB+ ideal).
 
-OpenHuman handles the rest: lifecycle (`src/openhuman/local_ai/service/`), API clients (`ollama_api.rs`, `lm_studio_api.rs`), health checks, and graceful fallback to remote when the local provider disappears.
+OpenHuman handles the rest: lifecycle (`src/openhuman/inference/local/service/`), API clients, health checks, and graceful fallback to remote when the local provider disappears.
 
 ### LM Studio troubleshooting
 
diff --git a/gitbooks/features/model-routing/local-ai.zh-CN.md b/gitbooks/features/model-routing/local-ai.zh-CN.md
index ccce11b31..1c577f8a8 100644
--- a/gitbooks/features/model-routing/local-ai.zh-CN.md
+++ b/gitbooks/features/model-routing/local-ai.zh-CN.md
@@ -65,7 +65,7 @@ OpenHuman 可以为以下工作负载在你机器上运行本地模型：当本
 | `local_ai.usage.learning_reflection` | `false` | 使用本地进行学习通过。 |
 | `local_ai.usage.subconscious` | `false` | 使用本地进行潜意识循环。 |
 
-在桌面 app 中，**设置 → AI 与技能 → 本地 AI** 暴露预设，选择一个（"仅嵌入向量"、"记忆 + 反思"、"全部本地"），正确的 flag 组合会为你设置。状态（Ollama 可达性、模型可用性、每个子系统启用）通过 `openhuman.local_ai_status` 实时暴露。
+在桌面 app 中，**设置 → AI 与技能 → 本地 AI** 暴露预设，选择一个（"仅嵌入向量"、"记忆 + 反思"、"全部本地"），正确的 flag 组合会为你设置。状态（Ollama 可达性、模型可用性、每个子系统启用）通过 `openhuman.inference_status` 实时暴露。
 
 ## 何时开启
 
@@ -83,7 +83,7 @@ OpenHuman 可以为以下工作负载在你机器上运行本地模型：当本
 * 模型有足够的磁盘（`gemma3:1b-it-qat` \~700 MB，`all-minilm:latest` \~23 MB）。
 * 有足够的 RAM 保持模型驻留（建议 8 GB+，理想 16 GB+）。
 
-OpenHuman 处理其余：生命周期（`src/openhuman/local_ai/service/`）、API 客户端（`ollama_api.rs`、`lm_studio_api.rs`）、健康检查，以及当本地 provider 消失时优雅地回退到远程。
+OpenHuman 处理其余：生命周期（`src/openhuman/inference/local/service/`）、API 客户端、健康检查，以及当本地 provider 消失时优雅地回退到远程。
 
 ### LM Studio 故障排除
 
@@ -96,4 +96,4 @@ OpenHuman 处理其余：生命周期（`src/openhuman/local_ai/service/`）、A
 
 * [记忆树](../obsidian-wiki/memory-tree.zh-CN.md)。本地嵌入向量 + 摘要 powering 什么。
 * [自动模型路由](README.zh-CN.md)。轻量聊天 hint 如何优先使用本地 provider。
-* [隐私与安全](../privacy-and-security.zh-CN.md)。当你 opt-in 时什么移至端侧。
\ No newline at end of file
+* [隐私与安全](../privacy-and-security.zh-CN.md)。当你 opt-in 时什么移至端侧。
diff --git a/src/core/all.rs b/src/core/all.rs
index 39e7391d0..15621e815 100644
--- a/src/core/all.rs
+++ b/src/core/all.rs
@@ -179,9 +179,9 @@ fn build_registered_controllers() -> Vec<RegisteredController> {
     // Background command monitors for agent-scoped event sources
     controllers.extend(crate::openhuman::monitor::all_monitor_registered_controllers());
     // Unified inference domain: text / vision / local runtime / cloud providers.
-    // (Formerly split across inference, local_ai, and providers namespaces.)
+    // (Formerly split across inference, local AI, and providers modules.)
     controllers.extend(crate::openhuman::inference::all_inference_registered_controllers());
-    controllers.extend(crate::openhuman::inference::all_local_ai_registered_controllers());
+    controllers.extend(crate::openhuman::inference::all_local_inference_registered_controllers());
     // Embedding provider configuration and embed RPC.
     controllers.extend(crate::openhuman::embeddings::all_embeddings_registered_controllers());
     // People resolution and interaction scoring
@@ -351,7 +351,7 @@ fn build_declared_controller_schemas() -> Vec<ControllerSchema> {
     schemas.extend(crate::openhuman::model_council::all_model_council_controller_schemas());
     schemas.extend(crate::openhuman::monitor::all_monitor_controller_schemas());
     schemas.extend(crate::openhuman::inference::all_inference_controller_schemas());
-    schemas.extend(crate::openhuman::inference::all_local_ai_controller_schemas());
+    schemas.extend(crate::openhuman::inference::all_local_inference_controller_schemas());
     schemas.extend(crate::openhuman::embeddings::all_embeddings_controller_schemas());
     schemas.extend(crate::openhuman::people::all_people_controller_schemas());
     schemas.extend(
@@ -469,7 +469,6 @@ pub fn namespace_description(namespace: &str) -> Option<&'static str> {
         "encrypt" => Some("Encrypt secure values managed by secret storage."),
         "health" => Some("Process and component health snapshots."),
         "inference" => Some("Connect to configured text, vision, and embedding inference runtimes."),
-        "local_ai" => Some("Local AI chat, inference, downloads, and media operations."),
         "migrate" => Some("Data migration utilities."),
         "javascript" => Some("First-class JavaScript runtime bridge for listing and dispatching tools."),
         "monitor" => Some("Start, inspect, read, and stop bounded background command monitors."),
diff --git a/src/core/cli_tests.rs b/src/core/cli_tests.rs
index 0b904e707..1dcf86e21 100644
--- a/src/core/cli_tests.rs
+++ b/src/core/cli_tests.rs
@@ -25,7 +25,6 @@ fn grouped_schemas_contains_migrated_namespaces() {
     assert!(grouped.contains_key("service"));
     assert!(grouped.contains_key("migrate"));
     assert!(grouped.contains_key("inference"));
-    assert!(grouped.contains_key("local_ai"));
 }
 
 #[test]
diff --git a/src/core/legacy_aliases.rs b/src/core/legacy_aliases.rs
index a712ee011..c059c9e15 100644
--- a/src/core/legacy_aliases.rs
+++ b/src/core/legacy_aliases.rs
@@ -111,6 +111,18 @@ const LEGACY_ALIASES: &[(&str, &str)] = &[
         "openhuman.local_ai_apply_preset",
         "openhuman.inference_apply_preset",
     ),
+    (
+        "openhuman.local_ai_agent_chat",
+        "openhuman.inference_agent_chat",
+    ),
+    (
+        "openhuman.local_ai_agent_chat_simple",
+        "openhuman.inference_agent_chat_simple",
+    ),
+    (
+        "openhuman.local_ai_assets_status",
+        "openhuman.inference_assets_status",
+    ),
     (
         "openhuman.local_ai_device_profile",
         "openhuman.inference_device_profile",
@@ -119,6 +131,26 @@ const LEGACY_ALIASES: &[(&str, &str)] = &[
         "openhuman.local_ai_diagnostics",
         "openhuman.inference_diagnostics",
     ),
+    (
+        "openhuman.local_ai_download_asset",
+        "openhuman.inference_download_asset",
+    ),
+    (
+        "openhuman.local_ai_downloads_progress",
+        "openhuman.inference_downloads_progress",
+    ),
+    (
+        "openhuman.local_ai_install_piper",
+        "openhuman.inference_install_piper",
+    ),
+    (
+        "openhuman.local_ai_install_whisper",
+        "openhuman.inference_install_whisper",
+    ),
+    (
+        "openhuman.local_ai_piper_install_status",
+        "openhuman.inference_piper_install_status",
+    ),
     // bare `health_snapshot` (no namespace prefix) was used by older clients
     // before the canonical `openhuman.health_snapshot` form was established.
     ("health_snapshot", "openhuman.health_snapshot"),
@@ -128,6 +160,23 @@ const LEGACY_ALIASES: &[(&str, &str)] = &[
     ("openhuman.system_info", "openhuman.health_system_info"),
     ("openhuman.inference_embed", "openhuman.embeddings_embed"),
     ("openhuman.local_ai_presets", "openhuman.inference_presets"),
+    (
+        "openhuman.local_ai_test_connection",
+        "openhuman.inference_test_connection",
+    ),
+    (
+        "openhuman.local_ai_transcribe",
+        "openhuman.inference_transcribe",
+    ),
+    (
+        "openhuman.local_ai_transcribe_bytes",
+        "openhuman.inference_transcribe_bytes",
+    ),
+    ("openhuman.local_ai_tts", "openhuman.inference_tts"),
+    (
+        "openhuman.local_ai_whisper_install_status",
+        "openhuman.inference_whisper_install_status",
+    ),
     (
         "openhuman.providers_list_models",
         "openhuman.inference_list_models",
diff --git a/src/core/observability.rs b/src/core/observability.rs
index c1d13735d..102e74476 100644
--- a/src/core/observability.rs
+++ b/src/core/observability.rs
@@ -1146,10 +1146,10 @@ fn is_provider_user_state_message(lower: &str) -> bool {
 /// The two canonical wire shapes today both contain `"for this ram tier"`:
 ///
 /// - `"Vision is disabled for this RAM tier. Switch to the 4-8 GB tier or
-///   above to enable it."` — from `local_ai/service/assets.rs::ensure_capability_ready`
+///   above to enable it."` — from `inference/local/service/assets.rs::ensure_capability_ready`
 /// - `"vision summaries are unavailable for this RAM tier. Use OCR-only
 ///   summarization or switch to a higher local AI tier."` —
-///   from `local_ai/service/vision_embed.rs::summarize`
+///   from `inference/local/service/vision_embed.rs::summarize`
 ///
 /// Anchor the classifier to that exact substring so an unrelated message
 /// that merely mentions "RAM tier" out of context is not silenced.
diff --git a/src/openhuman/agent/README.md b/src/openhuman/agent/README.md
index ceb9c3466..ea386a4b5 100644
--- a/src/openhuman/agent/README.md
+++ b/src/openhuman/agent/README.md
@@ -22,7 +22,7 @@ Multi-agent orchestration domain. Owns the LLM tool-calling loop, sub-agent disp
 - `src/openhuman/tools/` — `Tool` / `ToolSpec` execution surface invoked from the tool loop.
 - `src/openhuman/memory/` — episodic indexing + memory-loader context injection.
 - `src/openhuman/context/` — prompt sections, tool-call format selection.
-- `src/openhuman/local_ai/` — `agent_chat` / `agent_chat_simple` execution backend.
+- `src/openhuman/inference/local/` — `agent_chat` / `agent_chat_simple` execution backend.
 - `src/openhuman/config/` — runtime config load via `config::rpc::load_config_with_timeout`.
 - `src/core/event_bus/` — emits `DomainEvent::Agent(*)` and `Trigger*` events; subscribers in `agent/bus.rs`.
 
diff --git a/src/openhuman/agent/triage/evaluator.rs b/src/openhuman/agent/triage/evaluator.rs
index 774b81342..fe469717b 100644
--- a/src/openhuman/agent/triage/evaluator.rs
+++ b/src/openhuman/agent/triage/evaluator.rs
@@ -628,7 +628,7 @@ fn classify_error(message: String) -> ArmError {
 ///   `agent.run_turn`).
 /// - `src/openhuman/agent/harness/session/runtime.rs` — same strings in the
 ///   tool-call loop, kept identical so this classifier covers both.
-/// - `src/openhuman/local_ai/ops.rs` — user-facing variants with the
+/// - `src/openhuman/inference/local/ops.rs` — user-facing variants with the
 ///   `"Please rephrase clearly."` suffix; we match the leading phrase so
 ///   either form classifies.
 ///
diff --git a/src/openhuman/app_state/README.md b/src/openhuman/app_state/README.md
index b64891e83..04d85ae4c 100644
--- a/src/openhuman/app_state/README.md
+++ b/src/openhuman/app_state/README.md
@@ -17,7 +17,7 @@ Aggregator that the React shell polls every few seconds to render the OS-level c
 
 - `src/openhuman/config/` — `config_rpc::*` for `Config` reads and the workspace dir resolver.
 - `src/openhuman/autocomplete/` — `AutocompleteStatus` snapshot.
-- `src/openhuman/local_ai/` — `LocalAiStatus` snapshot.
+- `src/openhuman/inference/local/` — `LocalAiStatus` snapshot.
 - `src/openhuman/screen_intelligence/` — `AccessibilityStatus` snapshot.
 - `src/openhuman/service/` — `ServiceState` / `ServiceStatus` runtime info.
 - `src/openhuman/credentials/` — `session_support::build_session_state` for the auth slice.
diff --git a/src/openhuman/autocomplete/core/engine.rs b/src/openhuman/autocomplete/core/engine.rs
index 93a160a6a..8792e5f88 100644
--- a/src/openhuman/autocomplete/core/engine.rs
+++ b/src/openhuman/autocomplete/core/engine.rs
@@ -734,7 +734,7 @@ impl AutocompleteEngine {
         // Interactive variant — bypasses the scheduler_gate's LLM permit
         // so per-keystroke autocomplete doesn't queue behind a memory-tree
         // backfill or a triage turn. See `inline_complete_interactive`
-        // docs in `local_ai/service/public_infer.rs`.
+        // docs in `inference/local/service/public_infer.rs`.
         let generated = match service
             .inline_complete_interactive(
                 &config,
diff --git a/src/openhuman/config/README.md b/src/openhuman/config/README.md
index 1eb3b6a64..980f5313c 100644
--- a/src/openhuman/config/README.md
+++ b/src/openhuman/config/README.md
@@ -24,7 +24,7 @@ Authoritative TOML-backed configuration layer. Owns the `Config` schema (every d
 ## Called by
 
 - ~177 sites across the workspace — every domain pulls `Config` for its slice.
-- Hot consumers: `src/openhuman/agent/` (model + autonomy), `src/openhuman/channels/` (provider tokens), `src/openhuman/memory/` (storage paths), `src/openhuman/cron/` (scheduler poll), `src/openhuman/local_ai/` (Ollama / device routing), `src/openhuman/security/` (sandbox backend), `src/openhuman/voice/`, `src/openhuman/notifications/`, `src/openhuman/tools/`, `src/openhuman/encryption/`, `src/openhuman/tree_summarizer/`, `src/openhuman/referral/`.
+- Hot consumers: `src/openhuman/agent/` (model + autonomy), `src/openhuman/channels/` (provider tokens), `src/openhuman/memory/` (storage paths), `src/openhuman/cron/` (scheduler poll), `src/openhuman/inference/local/` (Ollama / device routing), `src/openhuman/security/` (sandbox backend), `src/openhuman/voice/`, `src/openhuman/notifications/`, `src/openhuman/tools/`, `src/openhuman/encryption/`, `src/openhuman/tree_summarizer/`, `src/openhuman/referral/`.
 - `src/core/all.rs` — registers `all_config_*`.
 
 ## Tests
diff --git a/src/openhuman/encryption/README.md b/src/openhuman/encryption/README.md
index 3c11dbb90..990fcaa3b 100644
--- a/src/openhuman/encryption/README.md
+++ b/src/openhuman/encryption/README.md
@@ -24,7 +24,7 @@ AES-256-GCM at-rest crypto for AI memory storage and the encrypt/decrypt RPC sur
 
 - `src/openhuman/credentials/` — uses the same `EncryptedPayload` / `EncryptionKey` primitives directly when storing per-channel secrets.
 - `src/core/all.rs` — registers `all_encryption_*` controllers so the shell + CLI can encrypt configuration secrets.
-- Indirect: `src/openhuman/memory/`, `src/openhuman/channels/`, and `src/openhuman/local_ai/` rely on the credentials domain (which in turn uses this layer) for secrets at rest.
+- Indirect: `src/openhuman/memory/`, `src/openhuman/channels/`, and `src/openhuman/inference/local/` rely on the credentials domain (which in turn uses this layer) for secrets at rest.
 
 ## Tests
 
diff --git a/src/openhuman/inference/README.md b/src/openhuman/inference/README.md
index d74e74bf6..3f8cf95e4 100644
--- a/src/openhuman/inference/README.md
+++ b/src/openhuman/inference/README.md
@@ -1,6 +1,6 @@
 # inference
 
-Unified inference domain: the canonical home for everything LLM/STT/TTS/embedding-related. It owns the local-runtime manager (Ollama / LM Studio / Whisper / Piper), the unified cloud + local provider abstraction (trait, factory, router, reliability/retry wrapper), voice transcription and TTS inference, OpenAI/Codex subscription OAuth, and an OpenAI-compatible `/v1/chat/completions` HTTP endpoint. It consolidates the previously separate `local_ai/`, `providers/`, and inference parts of `voice/` under one domain root. The RPC surface is still split across the `inference.*` and `local_ai.*` namespaces for backwards compatibility.
+Unified inference domain: the canonical home for everything LLM/STT/TTS/embedding-related. It owns the local-runtime manager (Ollama / LM Studio / Whisper / Piper), the unified cloud + local provider abstraction (trait, factory, router, reliability/retry wrapper), voice transcription and TTS inference, OpenAI/Codex subscription OAuth, and an OpenAI-compatible `/v1/chat/completions` HTTP endpoint. It consolidates the previously separate `local_ai/`, `providers/`, and inference parts of `voice/` under one domain root. The RPC surface is `inference.*`; older `local_ai_*` method names are compatibility aliases in `src/core/legacy_aliases.rs`.
 
 ## Responsibilities
 
@@ -33,7 +33,7 @@ Unified inference domain: the canonical home for everything LLM/STT/TTS/embeddin
 | `local/`                                                                          | Local runtime manager (was `local_ai/`).                                                                                                                                                                                                           |
 | `local/core.rs`                                                                   | `LocalAiService` singleton (`global`/`try_global`), `model_artifact_path`.                                                                                                                                                                         |
 | `local/ops.rs`                                                                    | Local RPC entrypoints (`local_ai_status/prompt/summarize/vision_prompt/embed/should_react`, `ReactionDecision`); re-exported as `local::rpc`.                                                                                                      |
-| `local/schemas.rs`                                                                | `local_ai.*` controller schemas + handlers.                                                                                                                                                                                                        |
+| `local/schemas.rs`                                                                | Local-runtime `inference.*` controller schemas + handlers.                                                                                                                                                                                         |
 | `local/ollama.rs`, `local/lm_studio.rs`                                           | Provider-specific runtime drivers; base-url resolution.                                                                                                                                                                                            |
 | `local/install*.rs`, `local/voice_install_common.rs`                              | Whisper/Piper install + shared download logic.                                                                                                                                                                                                     |
 | `local/model_requirements.rs`                                                     | `MIN_CONTEXT_TOKENS`, `evaluate_context`, `ContextEligibility`.                                                                                                                                                                                    |
@@ -65,18 +65,18 @@ From `mod.rs` re-exports:
 - `presets::{ModelPreset, ModelTier, VisionMode}`
 - `sentiment::SentimentResult`
 - `types::{LocalAiStatus, LocalAiAssetStatus, LocalAiAssetsStatus, LocalAiDownloadProgressItem, LocalAiDownloadsProgress, LocalAiEmbeddingResult, LocalAiSpeechResult, LocalAiTtsResult}`
-- `local::all_local_ai_controller_schemas` / `local::all_local_ai_registered_controllers`
+- `local::all_local_inference_controller_schemas` / `local::all_local_inference_registered_controllers` (legacy export names; registered schemas are in the `inference` namespace)
 - `rpc` (alias for `ops`) and `all_inference_controller_schemas` / `all_inference_registered_controllers`
 
 Provider-layer (via `provider::`): `Provider`, `ChatMessage`, `ChatRequest`, `ChatResponse`, `create_chat_provider`, `provider_for_role`, `BYOK_INCOMPLETE_SENTINEL`, plus error classifiers. Local runtime: `local::{global, try_global}` → `Arc<LocalAiService>`.
 
 ## RPC / controllers
 
-Two namespaces, both wired into the controller registry (`src/core/all.rs`).
+One namespace is wired into the controller registry (`src/core/all.rs`).
 
-`inference.*` (`schemas.rs`): `status`, `get_client_config`, `update_model_settings`, `update_local_settings`, `list_models`, `device_profile`, `presets`, `apply_preset`, `diagnostics`, `openai_oauth_start`, `openai_oauth_complete`, `openai_oauth_status`, `openai_oauth_disconnect`, `summarize`, `prompt`, `vision_prompt`, `test_provider_model`, `should_react`, `analyze_sentiment`.
+`inference.*` (`schemas.rs`, `local/schemas.rs`): `status`, `get_client_config`, `update_model_settings`, `update_local_settings`, `list_models`, `device_profile`, `presets`, `apply_preset`, `diagnostics`, `openai_oauth_start`, `openai_oauth_complete`, `openai_oauth_status`, `openai_oauth_disconnect`, `summarize`, `prompt`, `vision_prompt`, `test_provider_model`, `should_react`, `analyze_sentiment`, `agent_chat`, `agent_chat_simple`, `transcribe`, `transcribe_bytes`, `tts`, `assets_status`, `downloads_progress`, `download_asset`, `install_whisper`, `install_piper`, `whisper_install_status`, `piper_install_status`, `test_connection`.
 
-`local_ai.*` (`local/schemas.rs`): `agent_chat`, `agent_chat_simple`, `transcribe`, `transcribe_bytes`, `tts`, `assets_status`, `downloads_progress`, `download_asset`, `install_whisper`, `install_piper`, `whisper_install_status`, `piper_install_status`, `test_connection`.
+Legacy `openhuman.local_ai_*` and `openhuman.update_local_ai_settings` method names are rewritten to canonical `openhuman.inference_*` methods by `src/core/legacy_aliases.rs` and `app/src/services/rpcMethods.ts`.
 
 Also exposes a non-RPC HTTP router (`http::router()`) nested at `/v1` by `src/core/jsonrpc.rs` (`/v1/chat/completions`, `/v1/models`), accepting either the core bearer or a stable external API key.
 
diff --git a/src/openhuman/inference/local/mod.rs b/src/openhuman/inference/local/mod.rs
index 32296a8fd..0891f04d0 100644
--- a/src/openhuman/inference/local/mod.rs
+++ b/src/openhuman/inference/local/mod.rs
@@ -49,8 +49,8 @@ pub use core::*;
 pub use ops as rpc;
 pub use ops::*;
 pub use schemas::{
-    all_controller_schemas as all_local_ai_controller_schemas,
-    all_registered_controllers as all_local_ai_registered_controllers,
+    all_controller_schemas as all_local_inference_controller_schemas,
+    all_registered_controllers as all_local_inference_registered_controllers,
 };
 pub(crate) use service::whisper_engine;
 pub use service::LocalAiService;
diff --git a/src/openhuman/inference/local/schemas.rs b/src/openhuman/inference/local/schemas.rs
index cd961c18f..6d3db45ad 100644
--- a/src/openhuman/inference/local/schemas.rs
+++ b/src/openhuman/inference/local/schemas.rs
@@ -67,17 +67,17 @@ pub fn all_controller_schemas() -> Vec<ControllerSchema> {
     vec![
         schemas("agent_chat"),
         schemas("agent_chat_simple"),
-        schemas("local_ai_transcribe"),
-        schemas("local_ai_transcribe_bytes"),
-        schemas("local_ai_tts"),
-        schemas("local_ai_assets_status"),
-        schemas("local_ai_downloads_progress"),
-        schemas("local_ai_download_asset"),
-        schemas("local_ai_install_whisper"),
-        schemas("local_ai_install_piper"),
-        schemas("local_ai_whisper_install_status"),
-        schemas("local_ai_piper_install_status"),
-        schemas("local_ai_test_connection"),
+        schemas("transcribe"),
+        schemas("transcribe_bytes"),
+        schemas("tts"),
+        schemas("assets_status"),
+        schemas("downloads_progress"),
+        schemas("download_asset"),
+        schemas("install_whisper"),
+        schemas("install_piper"),
+        schemas("whisper_install_status"),
+        schemas("piper_install_status"),
+        schemas("test_connection"),
     ]
 }
 
@@ -92,47 +92,47 @@ pub fn all_registered_controllers() -> Vec<RegisteredController> {
             handler: handle_agent_chat_simple,
         },
         RegisteredController {
-            schema: schemas("local_ai_transcribe"),
+            schema: schemas("transcribe"),
             handler: handle_local_ai_transcribe,
         },
         RegisteredController {
-            schema: schemas("local_ai_transcribe_bytes"),
+            schema: schemas("transcribe_bytes"),
             handler: handle_local_ai_transcribe_bytes,
         },
         RegisteredController {
-            schema: schemas("local_ai_tts"),
+            schema: schemas("tts"),
             handler: handle_local_ai_tts,
         },
         RegisteredController {
-            schema: schemas("local_ai_assets_status"),
+            schema: schemas("assets_status"),
             handler: handle_local_ai_assets_status,
         },
         RegisteredController {
-            schema: schemas("local_ai_downloads_progress"),
+            schema: schemas("downloads_progress"),
             handler: handle_local_ai_downloads_progress,
         },
         RegisteredController {
-            schema: schemas("local_ai_download_asset"),
+            schema: schemas("download_asset"),
             handler: handle_local_ai_download_asset,
         },
         RegisteredController {
-            schema: schemas("local_ai_install_whisper"),
+            schema: schemas("install_whisper"),
             handler: handle_local_ai_install_whisper,
         },
         RegisteredController {
-            schema: schemas("local_ai_install_piper"),
+            schema: schemas("install_piper"),
             handler: handle_local_ai_install_piper,
         },
         RegisteredController {
-            schema: schemas("local_ai_whisper_install_status"),
+            schema: schemas("whisper_install_status"),
             handler: handle_local_ai_whisper_install_status,
         },
         RegisteredController {
-            schema: schemas("local_ai_piper_install_status"),
+            schema: schemas("piper_install_status"),
             handler: handle_local_ai_piper_install_status,
         },
         RegisteredController {
-            schema: schemas("local_ai_test_connection"),
+            schema: schemas("test_connection"),
             handler: handle_local_ai_test_connection,
         },
     ]
@@ -141,7 +141,7 @@ pub fn all_registered_controllers() -> Vec<RegisteredController> {
 pub fn schemas(function: &str) -> ControllerSchema {
     match function {
         "agent_chat" => ControllerSchema {
-            namespace: "local_ai",
+            namespace: "inference",
             function: "agent_chat",
             description: "Run one-shot agent chat with optional model overrides.",
             inputs: vec![
@@ -152,7 +152,7 @@ pub fn schemas(function: &str) -> ControllerSchema {
             outputs: vec![json_output("response", "Agent response payload.")],
         },
         "agent_chat_simple" => ControllerSchema {
-            namespace: "local_ai",
+            namespace: "inference",
             function: "agent_chat_simple",
             description: "Run one-shot lightweight provider chat.",
             inputs: vec![
@@ -162,15 +162,15 @@ pub fn schemas(function: &str) -> ControllerSchema {
             ],
             outputs: vec![json_output("response", "Agent response payload.")],
         },
-        "local_ai_transcribe" => ControllerSchema {
-            namespace: "local_ai",
+        "transcribe" => ControllerSchema {
+            namespace: "inference",
             function: "transcribe",
             description: "Transcribe audio from file path.",
             inputs: vec![required_string("audio_path", "Input audio path.")],
             outputs: vec![json_output("speech", "Transcription payload.")],
         },
-        "local_ai_transcribe_bytes" => ControllerSchema {
-            namespace: "local_ai",
+        "transcribe_bytes" => ControllerSchema {
+            namespace: "inference",
             function: "transcribe_bytes",
             description: "Transcribe audio from raw bytes.",
             inputs: vec![
@@ -184,8 +184,8 @@ pub fn schemas(function: &str) -> ControllerSchema {
             ],
             outputs: vec![json_output("speech", "Transcription payload.")],
         },
-        "local_ai_tts" => ControllerSchema {
-            namespace: "local_ai",
+        "tts" => ControllerSchema {
+            namespace: "inference",
             function: "tts",
             description: "Synthesize speech from text.",
             inputs: vec![
@@ -194,29 +194,29 @@ pub fn schemas(function: &str) -> ControllerSchema {
             ],
             outputs: vec![json_output("tts", "TTS result payload.")],
         },
-        "local_ai_assets_status" => ControllerSchema {
-            namespace: "local_ai",
+        "assets_status" => ControllerSchema {
+            namespace: "inference",
             function: "assets_status",
             description: "Get local AI asset installation status.",
             inputs: vec![],
             outputs: vec![json_output("status", "Assets status payload.")],
         },
-        "local_ai_downloads_progress" => ControllerSchema {
-            namespace: "local_ai",
+        "downloads_progress" => ControllerSchema {
+            namespace: "inference",
             function: "downloads_progress",
             description: "Get local AI download progress.",
             inputs: vec![],
             outputs: vec![json_output("progress", "Download progress payload.")],
         },
-        "local_ai_download_asset" => ControllerSchema {
-            namespace: "local_ai",
+        "download_asset" => ControllerSchema {
+            namespace: "inference",
             function: "download_asset",
             description: "Trigger download for one local AI asset capability.",
             inputs: vec![required_string("capability", "Asset capability id.")],
             outputs: vec![json_output("status", "Assets status payload.")],
         },
-        "local_ai_install_whisper" => ControllerSchema {
-            namespace: "local_ai",
+        "install_whisper" => ControllerSchema {
+            namespace: "inference",
             function: "install_whisper",
             description: "Download whisper.cpp's GGML model (and on Windows the whisper-cli binary) into the workspace so the local STT factory has everything it needs to run.",
             inputs: vec![
@@ -231,8 +231,8 @@ pub fn schemas(function: &str) -> ControllerSchema {
             ],
             outputs: vec![json_output("status", "Whisper install status payload.")],
         },
-        "local_ai_install_piper" => ControllerSchema {
-            namespace: "local_ai",
+        "install_piper" => ControllerSchema {
+            namespace: "inference",
             function: "install_piper",
             description: "Download the Piper binary archive and the bundled en_US-lessac-medium voice files into the workspace.",
             inputs: vec![
@@ -247,31 +247,31 @@ pub fn schemas(function: &str) -> ControllerSchema {
             ],
             outputs: vec![json_output("status", "Piper install status payload.")],
         },
-        "local_ai_whisper_install_status" => ControllerSchema {
-            namespace: "local_ai",
+        "whisper_install_status" => ControllerSchema {
+            namespace: "inference",
             function: "whisper_install_status",
             description: "Query the Whisper install state (missing / installing / installed / broken / error) plus per-stage download progress.",
             inputs: vec![],
             outputs: vec![json_output("status", "Whisper install status payload.")],
         },
-        "local_ai_piper_install_status" => ControllerSchema {
-            namespace: "local_ai",
+        "piper_install_status" => ControllerSchema {
+            namespace: "inference",
             function: "piper_install_status",
             description: "Query the Piper install state (missing / installing / installed / broken / error) plus per-stage download progress.",
             inputs: vec![],
             outputs: vec![json_output("status", "Piper install status payload.")],
         },
-        "local_ai_test_connection" => ControllerSchema {
-            namespace: "local_ai",
+        "test_connection" => ControllerSchema {
+            namespace: "inference",
             function: "test_connection",
             description: "Test connectivity to an Ollama server URL. Returns reachable status and model count.",
             inputs: vec![required_string("url", "Ollama server URL to test.")],
             outputs: vec![json_output("result", "Connection test result.")],
         },
         _ => ControllerSchema {
-            namespace: "local_ai",
+            namespace: "inference",
             function: "unknown",
-            description: "Unknown local_ai controller function.",
+            description: "Unknown local inference controller function.",
             inputs: vec![],
             outputs: vec![FieldSchema {
                 name: "error",
diff --git a/src/openhuman/inference/local/schemas_tests.rs b/src/openhuman/inference/local/schemas_tests.rs
index 8f7a6f329..94cefe6a1 100644
--- a/src/openhuman/inference/local/schemas_tests.rs
+++ b/src/openhuman/inference/local/schemas_tests.rs
@@ -5,13 +5,16 @@ fn catalog_counts_match_and_nonempty() {
     let s = all_controller_schemas();
     let h = all_registered_controllers();
     assert_eq!(s.len(), h.len());
-    assert!(s.len() >= 12, "local_ai should expose >=12 controller fns");
+    assert!(
+        s.len() >= 12,
+        "local inference should expose >=12 controller fns"
+    );
 }
 
 #[test]
-fn all_schemas_use_local_ai_namespace_and_have_descriptions() {
+fn all_schemas_use_inference_namespace_and_have_descriptions() {
     for s in all_controller_schemas() {
-        assert_eq!(s.namespace, "local_ai", "function {}", s.function);
+        assert_eq!(s.namespace, "inference", "function {}", s.function);
         assert!(!s.description.is_empty(), "function {} desc", s.function);
         assert!(!s.outputs.is_empty(), "function {} outputs", s.function);
     }
@@ -21,7 +24,7 @@ fn all_schemas_use_local_ai_namespace_and_have_descriptions() {
 fn unknown_function_returns_unknown_schema() {
     let s = schemas("no_such_fn");
     assert_eq!(s.function, "unknown");
-    assert_eq!(s.namespace, "local_ai");
+    assert_eq!(s.namespace, "inference");
 }
 
 #[test]
@@ -29,29 +32,29 @@ fn every_registered_key_resolves_to_non_unknown_schema() {
     let keys = [
         "agent_chat",
         "agent_chat_simple",
-        "local_ai_transcribe",
-        "local_ai_transcribe_bytes",
-        "local_ai_tts",
-        "local_ai_assets_status",
-        "local_ai_downloads_progress",
-        "local_ai_download_asset",
-        "local_ai_install_whisper",
-        "local_ai_install_piper",
-        "local_ai_whisper_install_status",
-        "local_ai_piper_install_status",
-        "local_ai_test_connection",
+        "transcribe",
+        "transcribe_bytes",
+        "tts",
+        "assets_status",
+        "downloads_progress",
+        "download_asset",
+        "install_whisper",
+        "install_piper",
+        "whisper_install_status",
+        "piper_install_status",
+        "test_connection",
     ];
     for k in keys {
         let s = schemas(k);
-        assert_eq!(s.namespace, "local_ai");
+        assert_eq!(s.namespace, "inference");
         assert_ne!(s.function, "unknown", "key `{k}` fell through");
     }
 }
 
 #[test]
-fn registered_controllers_all_in_local_ai_namespace() {
+fn registered_controllers_all_in_inference_namespace() {
     for h in all_registered_controllers() {
-        assert_eq!(h.schema.namespace, "local_ai");
+        assert_eq!(h.schema.namespace, "inference");
         assert!(!h.schema.function.is_empty());
     }
 }
diff --git a/src/openhuman/inference/mod.rs b/src/openhuman/inference/mod.rs
index 7a04d75ec..b0c4bfb29 100644
--- a/src/openhuman/inference/mod.rs
+++ b/src/openhuman/inference/mod.rs
@@ -9,8 +9,8 @@
 //!                 (moved from `src/openhuman/voice/`)
 //! - `http/`     — OpenAI-compatible `/v1/chat/completions` endpoint
 //!
-//! The RPC surface remains under the `inference.*` and `local_ai.*` namespaces
-//! for backwards compatibility.
+//! The RPC surface is `inference.*`; old `local_ai_*` RPC names are resolved
+//! by the legacy alias layer for backwards compatibility.
 
 pub mod device;
 pub mod http;
@@ -36,8 +36,8 @@ pub use schemas::{
 
 // Re-export the types that external callers (voice, agent, etc.) import from inference
 pub use device::DeviceProfile;
-pub use local::all_local_ai_controller_schemas;
-pub use local::all_local_ai_registered_controllers;
+pub use local::all_local_inference_controller_schemas;
+pub use local::all_local_inference_registered_controllers;
 pub use model_context::context_window_for_model;
 pub use presets::{ModelPreset, ModelTier, VisionMode};
 pub use sentiment::SentimentResult;
diff --git a/src/openhuman/inference/voice/local_speech.rs b/src/openhuman/inference/voice/local_speech.rs
index 3c966daac..ee41f2146 100644
--- a/src/openhuman/inference/voice/local_speech.rs
+++ b/src/openhuman/inference/voice/local_speech.rs
@@ -39,7 +39,7 @@
 //! into `~/.openhuman/bin/piper/`, extracts it, and stages the bundled
 //! `en_US-lessac-medium` voice (`.onnx` + `.onnx.json`) alongside via a
 //! `.part` file + atomic rename. After install the `resolve_piper_binary`
-//! helper in `local_ai/paths.rs` picks it up automatically.
+//! helper in `inference/paths.rs` picks it up automatically.
 //!
 //! **Advanced path:** download Piper from
 //! [rhasspy/piper](https://github.com/rhasspy/piper) releases (one
diff --git a/src/openhuman/inference/voice/local_transcribe.rs b/src/openhuman/inference/voice/local_transcribe.rs
index 7822b4f74..ecf6899dc 100644
--- a/src/openhuman/inference/voice/local_transcribe.rs
+++ b/src/openhuman/inference/voice/local_transcribe.rs
@@ -20,7 +20,7 @@
 //! `~/.openhuman/bin/whisper/` via a `.part` file + atomic rename, plus
 //! the `whisper-cli` binary on Windows where upstream ships a release
 //! archive. After install the `resolve_whisper_binary` helper in
-//! `local_ai/paths.rs` picks it up automatically — no env var to set.
+//! `inference/paths.rs` picks it up automatically — no env var to set.
 //!
 //! **Advanced path:** install whisper.cpp's `whisper-cli` from a package
 //! manager (`brew install whisper-cpp`, `pacman -S whisper.cpp`, …) or
diff --git a/tests/inference_agent_raw_coverage_e2e.rs b/tests/inference_agent_raw_coverage_e2e.rs
index eb9f0121f..48083e495 100644
--- a/tests/inference_agent_raw_coverage_e2e.rs
+++ b/tests/inference_agent_raw_coverage_e2e.rs
@@ -172,7 +172,8 @@ use openhuman_core::openhuman::inference::voice::local_speech::{synthesize_piper
 use openhuman_core::openhuman::inference::voice::postprocess::cleanup_transcription;
 use openhuman_core::openhuman::inference::{
     all_inference_controller_schemas, all_inference_registered_controllers,
-    all_local_ai_controller_schemas, all_local_ai_registered_controllers, DeviceProfile,
+    all_local_inference_controller_schemas, all_local_inference_registered_controllers,
+    DeviceProfile,
 };
 use openhuman_core::openhuman::memory::{Memory, MemoryCategory, MemoryEntry, RecallOpts};
 use openhuman_core::openhuman::security::SecurityPolicy;
@@ -3018,13 +3019,13 @@ async fn inference_local_controllers_and_presets_cover_public_paths() {
     let _ollama_bin_guard = EnvVarGuard::set("OLLAMA_BIN", &mock_ollama);
     let _ollama_base_guard = EnvVarGuard::set("OPENHUMAN_OLLAMA_BASE_URL", &provider_base);
 
-    let local_schemas = all_local_ai_controller_schemas();
-    let local_registered = all_local_ai_registered_controllers();
+    let local_schemas = all_local_inference_controller_schemas();
+    let local_registered = all_local_inference_registered_controllers();
     assert_eq!(local_schemas.len(), local_registered.len());
     assert!(local_registered.iter().all(|controller| {
         controller
             .rpc_method_name()
-            .starts_with("openhuman.local_ai_")
+            .starts_with("openhuman.inference_")
     }));
 
     let reachable = call(
diff --git a/tests/inference_compatible_admin_round25_raw_coverage_e2e.rs b/tests/inference_compatible_admin_round25_raw_coverage_e2e.rs
index 062794001..93cdc7907 100644
--- a/tests/inference_compatible_admin_round25_raw_coverage_e2e.rs
+++ b/tests/inference_compatible_admin_round25_raw_coverage_e2e.rs
@@ -20,7 +20,7 @@ use openhuman_core::openhuman::config::schema::cloud_providers::{
 };
 use openhuman_core::openhuman::config::Config;
 use openhuman_core::openhuman::credentials::{AuthService, DEFAULT_AUTH_PROFILE_NAME};
-use openhuman_core::openhuman::inference::local::all_local_ai_registered_controllers;
+use openhuman_core::openhuman::inference::local::all_local_inference_registered_controllers;
 use openhuman_core::openhuman::inference::ops::inference_test_provider_model;
 use openhuman_core::openhuman::inference::provider::compatible::{
     AuthStyle as CompatibleAuthStyle, OpenAiCompatibleProvider,
@@ -226,7 +226,7 @@ async fn provider_admin_cold_paths_cover_model_errors_local_factory_and_connecti
     .expect_err("empty lmstudio model");
     assert!(empty_lmstudio.contains("empty model"));
 
-    let controllers = all_local_ai_registered_controllers();
+    let controllers = all_local_inference_registered_controllers();
     let test_connection = controller(&controllers, "test_connection");
     let reachable = call(test_connection, json!({"url": base}))
         .await
diff --git a/tests/inference_local_ops_piper_raw_coverage_e2e.rs b/tests/inference_local_ops_piper_raw_coverage_e2e.rs
index fec77897f..ebc356ca0 100644
--- a/tests/inference_local_ops_piper_raw_coverage_e2e.rs
+++ b/tests/inference_local_ops_piper_raw_coverage_e2e.rs
@@ -18,7 +18,7 @@ use flate2::Compression;
 use openhuman_core::core::all::RegisteredController;
 use openhuman_core::openhuman::config::Config;
 use openhuman_core::openhuman::inference::local::{
-    all_local_ai_registered_controllers, local_ai_transcribe_bytes,
+    all_local_inference_registered_controllers, local_ai_transcribe_bytes,
 };
 use serde_json::{json, Value};
 use tempfile::{tempdir, TempDir};
@@ -107,7 +107,7 @@ async fn piper_controller_installs_skips_existing_and_records_failures_from_mock
     let _piper_bin = EnvVarGuard::unset("PIPER_BIN");
     let _whisper_bin = EnvVarGuard::unset("WHISPER_BIN");
 
-    let controllers = all_local_ai_registered_controllers();
+    let controllers = all_local_inference_registered_controllers();
     let install = controller(&controllers, "install_piper");
     let status = controller(&controllers, "piper_install_status");
 
diff --git a/tests/inference_local_services_round21_raw_coverage_e2e.rs b/tests/inference_local_services_round21_raw_coverage_e2e.rs
index b451f622c..918044ac4 100644
--- a/tests/inference_local_services_round21_raw_coverage_e2e.rs
+++ b/tests/inference_local_services_round21_raw_coverage_e2e.rs
@@ -21,7 +21,7 @@ use openhuman_core::openhuman::inference::local::ops::{
     LocalAiChatMessage,
 };
 use openhuman_core::openhuman::inference::local::{
-    all_local_ai_registered_controllers, LocalAiService,
+    all_local_inference_registered_controllers, LocalAiService,
 };
 use serde_json::{json, Value};
 use tempfile::{tempdir, TempDir};
@@ -295,7 +295,7 @@ async fn local_services_cover_mocked_inference_assets_speech_and_whisper_install
         "ready"
     );
 
-    let controllers = all_local_ai_registered_controllers();
+    let controllers = all_local_inference_registered_controllers();
     let install = controller(&controllers, "install_whisper");
     let status = controller(&controllers, "whisper_install_status");
 
diff --git a/tests/json_rpc_e2e.rs b/tests/json_rpc_e2e.rs
index b463c4fbd..22efd5b2a 100644
--- a/tests/json_rpc_e2e.rs
+++ b/tests/json_rpc_e2e.rs
@@ -1764,7 +1764,7 @@ async fn json_rpc_protocol_auth_and_agent_hello_inner() {
     let chat = post_json_rpc(
         &rpc_base,
         5,
-        "openhuman.local_ai_agent_chat",
+        "openhuman.inference_agent_chat",
         json!({
             "message": "Hello",
         }),
@@ -2207,14 +2207,14 @@ async fn json_rpc_prompt_injection_is_rejected_before_model_call() {
     let blocked_agent = post_json_rpc(
         &rpc_base,
         4003,
-        "openhuman.local_ai_agent_chat",
+        "openhuman.inference_agent_chat",
         json!({
             "message": payload,
             "model_override": "e2e-mock-model",
         }),
     )
     .await;
-    let agent_err = assert_jsonrpc_error(&blocked_agent, "local_ai_agent_chat blocked");
+    let agent_err = assert_jsonrpc_error(&blocked_agent, "inference_agent_chat blocked");
     let agent_msg = agent_err
         .get("message")
         .and_then(Value::as_str)