diff --git a/api/models.py b/api/models.py
index 0c96543..80261b6 100644
--- a/api/models.py
+++ b/api/models.py
@@ -395,6 +395,16 @@ class SourceInsightResponse(BaseModel):
     updated: str
 
 
+class InsightCreationResponse(BaseModel):
+    """Response for async insight creation."""
+
+    status: Literal["pending"] = "pending"
+    message: str = "Insight generation started"
+    source_id: str
+    transformation_id: str
+    command_id: Optional[str] = None
+
+
 class SaveAsNoteRequest(BaseModel):
     notebook_id: Optional[str] = Field(None, description="Notebook ID to add note to")
 
diff --git a/api/routers/sources.py b/api/routers/sources.py
index 22e4f74..e2cd62d 100644
--- a/api/routers/sources.py
+++ b/api/routers/sources.py
@@ -14,12 +14,13 @@ from fastapi import (
 )
 from fastapi.responses import FileResponse, Response
 from loguru import logger
-from surreal_commands import execute_command_sync
+from surreal_commands import execute_command_sync, submit_command
 
 from api.command_service import CommandService
 from api.models import (
     AssetModel,
     CreateSourceInsightRequest,
+    InsightCreationResponse,
     SourceCreate,
     SourceInsightResponse,
     SourceListResponse,
@@ -963,44 +964,56 @@ async def get_source_insights(source_id: str):
         )
 
 
-@router.post("/sources/{source_id}/insights", response_model=SourceInsightResponse)
+@router.post(
+    "/sources/{source_id}/insights",
+    response_model=InsightCreationResponse,
+    status_code=202,
+)
 async def create_source_insight(source_id: str, request: CreateSourceInsightRequest):
-    """Create a new insight for a source by running a transformation."""
+    """
+    Start insight generation for a source by running a transformation.
+
+    This endpoint returns immediately with a 202 Accepted status.
+    The transformation runs asynchronously in the background via the job queue.
+    Poll GET /sources/{source_id}/insights to see when the insight is ready.
+    """
     try:
-        # Get source
+        # Validate source exists
         source = await Source.get(source_id)
         if not source:
             raise HTTPException(status_code=404, detail="Source not found")
 
-        # Get transformation
+        # Validate transformation exists
         transformation = await Transformation.get(request.transformation_id)
         if not transformation:
             raise HTTPException(status_code=404, detail="Transformation not found")
 
-        # Run transformation graph
-        from open_notebook.graphs.transformation import graph as transform_graph
-
-        await transform_graph.ainvoke(
-            input=dict(source=source, transformation=transformation)  # type: ignore[arg-type]
+        # Submit transformation as background job (fire-and-forget)
+        command_id = submit_command(
+            "open_notebook",
+            "run_transformation",
+            {
+                "source_id": source_id,
+                "transformation_id": request.transformation_id,
+            },
+        )
+        logger.info(
+            f"Submitted run_transformation command {command_id} for source {source_id}"
         )
 
-        # Get the newly created insight (last one)
-        insights = await source.get_insights()
-        if insights:
-            newest = insights[-1]
-            return SourceInsightResponse(
-                id=newest.id or "",
-                source_id=source_id,
-                insight_type=newest.insight_type,
-                content=newest.content,
-                created=str(newest.created),
-                updated=str(newest.updated),
-            )
-        else:
-            raise HTTPException(status_code=500, detail="Failed to create insight")
+        # Return immediately with command_id for status tracking
+        return InsightCreationResponse(
+            status="pending",
+            message="Insight generation started",
+            source_id=source_id,
+            transformation_id=request.transformation_id,
+            command_id=str(command_id),
+        )
 
     except HTTPException:
         raise
     except Exception as e:
-        logger.error(f"Error creating insight for source {source_id}: {str(e)}")
-        raise HTTPException(status_code=500, detail=f"Error creating insight: {str(e)}")
+        logger.error(f"Error starting insight generation for source {source_id}: {e}")
+        raise HTTPException(
+            status_code=500, detail=f"Error starting insight generation: {str(e)}"
+        )
diff --git a/commands/CLAUDE.md b/commands/CLAUDE.md
index eb9174e..5fe6f2f 100644
--- a/commands/CLAUDE.md
+++ b/commands/CLAUDE.md
@@ -9,11 +9,13 @@
 - **`embed_note_command`**: Embeds a single note using unified embedding pipeline with content-type aware processing. Uses MARKDOWN content type detection. Retry: 5 attempts, exponential jitter 1-60s.
 - **`embed_insight_command`**: Embeds a single source insight. Uses MARKDOWN content type. Retry: 5 attempts, exponential jitter 1-60s.
 - **`embed_source_command`**: Embeds a source by chunking full_text with content-type aware splitters (HTML, Markdown, plain), then batch embedding all chunks. Uses single Esperanto API call. Retry: 5 attempts, exponential jitter 1-60s.
+- **`create_insight_command`**: Creates a source insight with automatic retry on transaction conflicts. Creates the DB record, then submits `embed_insight` command (fire-and-forget). Retry: 5 attempts, exponential jitter 1-60s. Used by `Source.add_insight()`.
 - **`rebuild_embeddings_command`**: Submits individual embed_* commands for all sources/notes/insights. Returns immediately; actual embedding happens async. No retry (coordinator only).
 
 ### Other Commands
 
-- **`process_source_command`**: Ingests content through `source_graph`, creates embeddings (optional), and generates insights. Retries on transaction conflicts (exp. jitter, max 5×).
+- **`process_source_command`**: Ingests content through `source_graph`, creates embeddings (optional), and generates insights. Retries on transaction conflicts (exp. jitter, max 15×, 1-120s).
+- **`run_transformation_command`**: Runs a transformation on an existing source to generate an insight. Executes the transformation graph (LLM call) then creates insight via `create_insight_command`. Used by `POST /sources/{id}/insights` API endpoint. Retry: 5 attempts, exponential jitter 1-60s.
 - **`generate_podcast_command`**: Creates podcasts via `podcast-creator` library using stored episode/speaker profiles.
 - **`process_text_command`** (example): Test fixture for text operations (uppercase, lowercase, reverse, word_count).
 - **`analyze_data_command`** (example): Test fixture for numeric aggregations.
diff --git a/commands/embedding_commands.py b/commands/embedding_commands.py
index f5e05ad..1087066 100644
--- a/commands/embedding_commands.py
+++ b/commands/embedding_commands.py
@@ -47,6 +47,23 @@ class RebuildEmbeddingsOutput(CommandOutput):
 # =============================================================================
 
 
+class CreateInsightInput(CommandInput):
+    """Input for creating a source insight with automatic retry on conflicts."""
+
+    source_id: str
+    insight_type: str
+    content: str
+
+
+class CreateInsightOutput(CommandOutput):
+    """Output from insight creation command."""
+
+    success: bool
+    insight_id: Optional[str] = None
+    processing_time: float
+    error_message: Optional[str] = None
+
+
 class EmbedNoteInput(CommandInput):
     """Input for embedding a single note."""
 
@@ -412,6 +429,115 @@ async def embed_source_command(input_data: EmbedSourceInput) -> EmbedSourceOutpu
         )
 
 
+@command(
+    "create_insight",
+    app="open_notebook",
+    retry={
+        "max_attempts": 5,
+        "wait_strategy": "exponential_jitter",
+        "wait_min": 1,
+        "wait_max": 60,
+        "retry_on": [RuntimeError, ConnectionError, TimeoutError],
+        "retry_log_level": "debug",
+    },
+)
+async def create_insight_command(
+    input_data: CreateInsightInput,
+) -> CreateInsightOutput:
+    """
+    Create a source insight with automatic retry on transaction conflicts.
+
+    This command wraps the CREATE source_insight operation with retry logic
+    to handle SurrealDB transaction conflicts that occur during batch imports
+    when multiple parallel transformations try to create insights concurrently.
+
+    Flow:
+    1. CREATE source_insight record in database
+    2. Submit embed_insight command (fire-and-forget) for async embedding
+    3. Return the insight_id
+
+    Retry Strategy:
+    - Retries up to 5 times for transient failures (RuntimeError, ConnectionError, TimeoutError)
+    - Uses exponential-jitter backoff (1-60s)
+    - Does NOT retry permanent failures (ValueError, authentication errors)
+    """
+    start_time = time.time()
+
+    try:
+        logger.info(
+            f"Creating insight for source {input_data.source_id}: "
+            f"type={input_data.insight_type}"
+        )
+
+        # 1. Create insight record in database
+        result = await repo_query(
+            """
+            CREATE source_insight CONTENT {
+                "source": $source_id,
+                "insight_type": $insight_type,
+                "content": $content
+            };
+            """,
+            {
+                "source_id": ensure_record_id(input_data.source_id),
+                "insight_type": input_data.insight_type,
+                "content": input_data.content,
+            },
+        )
+
+        if not result or len(result) == 0:
+            raise ValueError("Failed to create insight - no result returned")
+
+        insight_id = str(result[0].get("id", ""))
+        if not insight_id:
+            raise ValueError("Failed to create insight - no ID in result")
+
+        # 2. Submit embedding command (fire-and-forget)
+        submit_command(
+            "open_notebook",
+            "embed_insight",
+            {"insight_id": insight_id},
+        )
+        logger.debug(f"Submitted embed_insight command for {insight_id}")
+
+        processing_time = time.time() - start_time
+        logger.info(
+            f"Successfully created insight {insight_id} for source "
+            f"{input_data.source_id} in {processing_time:.2f}s"
+        )
+
+        return CreateInsightOutput(
+            success=True,
+            insight_id=insight_id,
+            processing_time=processing_time,
+        )
+
+    except RuntimeError:
+        logger.debug(
+            f"Transaction conflict creating insight for source "
+            f"{input_data.source_id} - will retry"
+        )
+        raise
+    except (ConnectionError, TimeoutError) as e:
+        logger.debug(
+            f"Network/timeout error creating insight for source "
+            f"{input_data.source_id} ({type(e).__name__}: {e}) - will retry"
+        )
+        raise
+    except Exception as e:
+        processing_time = time.time() - start_time
+        logger.error(
+            f"Failed to create insight for source {input_data.source_id}: {e}"
+        )
+        logger.exception(e)
+
+        return CreateInsightOutput(
+            success=False,
+            processing_time=processing_time,
+            error_message=str(e),
+        )
+
+
 async def collect_items_for_rebuild(
     mode: str,
     include_sources: bool,
diff --git a/commands/source_commands.py b/commands/source_commands.py
index 0b42c44..b5fde95 100644
--- a/commands/source_commands.py
+++ b/commands/source_commands.py
@@ -11,9 +11,10 @@ from open_notebook.domain.transformation import Transformation
 
 try:
     from open_notebook.graphs.source import source_graph
+    from open_notebook.graphs.transformation import graph as transform_graph
 except ImportError as e:
-    logger.error(f"Failed to import source_graph: {e}")
-    raise ValueError("source_graph not available")
+    logger.error(f"Failed to import graphs: {e}")
+    raise ValueError("graphs not available")
 
 
 def full_model_dump(model):
@@ -151,3 +152,121 @@ async def process_source_command(
             processing_time=processing_time,
             error_message=str(e),
         )
+
+
+# =============================================================================
+# RUN TRANSFORMATION COMMAND
+# =============================================================================
+
+
+class RunTransformationInput(CommandInput):
+    """Input for running a transformation on an existing source."""
+
+    source_id: str
+    transformation_id: str
+
+
+class RunTransformationOutput(CommandOutput):
+    """Output from transformation command."""
+
+    success: bool
+    source_id: str
+    transformation_id: str
+    processing_time: float
+    error_message: Optional[str] = None
+
+
+@command(
+    "run_transformation",
+    app="open_notebook",
+    retry={
+        "max_attempts": 5,
+        "wait_strategy": "exponential_jitter",
+        "wait_min": 1,
+        "wait_max": 60,
+        "retry_on": [RuntimeError, ConnectionError, TimeoutError],
+        "retry_log_level": "debug",
+    },
+)
+async def run_transformation_command(
+    input_data: RunTransformationInput,
+) -> RunTransformationOutput:
+    """
+    Run a transformation on an existing source to generate an insight.
+
+    This command runs the transformation graph which:
+    1. Loads the source and transformation
+    2. Calls the LLM to generate insight content
+    3. Creates the insight via create_insight command (fire-and-forget)
+
+    Use this command for UI-triggered insight generation to avoid blocking
+    the HTTP request while the LLM processes.
+
+    Retry Strategy:
+    - Retries up to 5 times for transient failures
+    - Uses exponential-jitter backoff (1-60s)
+    """
+    start_time = time.time()
+
+    try:
+        logger.info(
+            f"Running transformation {input_data.transformation_id} "
+            f"on source {input_data.source_id}"
+        )
+
+        # Load source
+        source = await Source.get(input_data.source_id)
+        if not source:
+            raise ValueError(f"Source '{input_data.source_id}' not found")
+
+        # Load transformation
+        transformation = await Transformation.get(input_data.transformation_id)
+        if not transformation:
+            raise ValueError(
+                f"Transformation '{input_data.transformation_id}' not found"
+            )
+
+        # Run transformation graph (includes LLM call + insight creation)
+        await transform_graph.ainvoke(
+            input=dict(source=source, transformation=transformation)
+        )
+
+        processing_time = time.time() - start_time
+        logger.info(
+            f"Successfully ran transformation {input_data.transformation_id} "
+            f"on source {input_data.source_id} in {processing_time:.2f}s"
+        )
+
+        return RunTransformationOutput(
+            success=True,
+            source_id=input_data.source_id,
+            transformation_id=input_data.transformation_id,
+            processing_time=processing_time,
+        )
+
+    except RuntimeError:
+        logger.debug(
+            f"Transaction conflict running transformation - will retry"
+        )
+        raise
+    except (ConnectionError, TimeoutError) as e:
+        logger.debug(
+            f"Network/timeout error running transformation "
+            f"({type(e).__name__}: {e}) - will retry"
+        )
+        raise
+    except Exception as e:
+        processing_time = time.time() - start_time
+        logger.error(
+            f"Failed to run transformation {input_data.transformation_id} "
+            f"on source {input_data.source_id}: {e}"
+        )
+        logger.exception(e)
+
+        return RunTransformationOutput(
+            success=False,
+            source_id=input_data.source_id,
+            transformation_id=input_data.transformation_id,
+            processing_time=processing_time,
+            error_message=str(e),
+        )
diff --git a/frontend/src/app/(dashboard)/notebooks/[id]/page.tsx b/frontend/src/app/(dashboard)/notebooks/[id]/page.tsx
index 33e1d21..c942bc5 100644
--- a/frontend/src/app/(dashboard)/notebooks/[id]/page.tsx
+++ b/frontend/src/app/(dashboard)/notebooks/[id]/page.tsx
@@ -58,16 +58,21 @@ export default function NotebookPage() {
     notes: {}
   })
 
-  // Initialize default selections when sources/notes load
+  // Initialize and update selections when sources load or change
   useEffect(() => {
     if (sources && sources.length > 0) {
       setContextSelections(prev => {
         const newSourceSelections = { ...prev.sources }
         sources.forEach(source => {
-          // Only set default if not already set
-          if (!(source.id in newSourceSelections)) {
-            // Default to 'insights' if has insights, otherwise 'full'
-            newSourceSelections[source.id] = source.insights_count > 0 ? 'insights' : 'full'
+          const currentMode = newSourceSelections[source.id]
+          const hasInsights = source.insights_count > 0
+
+          if (currentMode === undefined) {
+            // Initial setup - default based on insights availability
+            newSourceSelections[source.id] = hasInsights ? 'insights' : 'full'
+          } else if (currentMode === 'full' && hasInsights) {
+            // Source gained insights while in 'full' mode - auto-switch to 'insights'
+            newSourceSelections[source.id] = 'insights'
           }
         })
         return { ...prev, sources: newSourceSelections }
diff --git a/frontend/src/components/source/SourceDetailContent.tsx b/frontend/src/components/source/SourceDetailContent.tsx
index a17c33d..87ebb40 100644
--- a/frontend/src/components/source/SourceDetailContent.tsx
+++ b/frontend/src/components/source/SourceDetailContent.tsx
@@ -1,6 +1,7 @@
 'use client'
 
 import { useState, useEffect, useCallback, useMemo } from 'react'
+import { useQueryClient } from '@tanstack/react-query'
 import { isAxiosError } from 'axios'
 import ReactMarkdown from 'react-markdown'
 import remarkGfm from 'remark-gfm'
@@ -81,6 +82,7 @@ export function SourceDetailContent({
   onClose
 }: SourceDetailContentProps) {
   const { t, language } = useTranslation()
+  const queryClient = useQueryClient()
   const [source, setSource] = useState<SourceDetailResponse | null>(null)
   const [insights, setInsights] = useState<SourceInsightResponse[]>([])
   const [transformations, setTransformations] = useState<Transformation[]>([])
@@ -154,12 +156,36 @@ export function SourceDetailContent({
 
     try {
       setCreatingInsight(true)
-      await insightsApi.create(sourceId, {
+      const response = await insightsApi.create(sourceId, {
         transformation_id: selectedTransformation
       })
-      toast.success(t.common.success)
-      await fetchInsights()
+      // Show toast for async operation
+      toast.success(t.sources.insightGenerationStarted)
       setSelectedTransformation('')
+
+      // Poll for command completion if we have a command_id
+      if (response.command_id) {
+        // Poll in background (don't block UI)
+        insightsApi.waitForCommand(response.command_id, {
+          maxAttempts: 120, // Up to 4 minutes (120 * 2s)
+          intervalMs: 2000
+        }).then(success => {
+          if (success) {
+            void fetchInsights()
+            // Invalidate sources queries so notebook page refreshes with updated insights_count
+            queryClient.invalidateQueries({ queryKey: ['sources'] })
+          }
+        }).catch(err => {
+          console.error('Error waiting for insight command:', err)
+        })
+      } else {
+        // Fallback: refresh after delay if no command_id
+        setTimeout(() => {
+          void fetchInsights()
+          // Also invalidate sources queries
+          queryClient.invalidateQueries({ queryKey: ['sources'] })
+        }, 5000)
+      }
     } catch (err) {
       console.error('Failed to create insight:', err)
       toast.error(t.common.error)
diff --git a/frontend/src/lib/api/insights.ts b/frontend/src/lib/api/insights.ts
index 1f6677c..ef5fd90 100644
--- a/frontend/src/lib/api/insights.ts
+++ b/frontend/src/lib/api/insights.ts
@@ -13,6 +13,21 @@ export interface CreateSourceInsightRequest {
   transformation_id: string
 }
 
+export interface InsightCreationResponse {
+  status: 'pending'
+  message: string
+  source_id: string
+  transformation_id: string
+  command_id?: string
+}
+
+export interface CommandJobStatusResponse {
+  job_id: string
+  status: string
+  result?: Record<string, unknown>
+  error_message?: string
+}
+
 export const insightsApi = {
   listForSource: async (sourceId: string) => {
     const response = await apiClient.get<SourceInsightResponse[]>(`/sources/${sourceId}/insights`)
@@ -25,7 +40,7 @@ export const insightsApi = {
   },
 
   create: async (sourceId: string, data: CreateSourceInsightRequest) => {
-    const response = await apiClient.post<SourceInsightResponse>(
+    const response = await apiClient.post<InsightCreationResponse>(
       `/sources/${sourceId}/insights`,
       data
     )
@@ -34,5 +49,46 @@ export const insightsApi = {
 
   delete: async (insightId: string) => {
     await apiClient.delete(`/insights/${insightId}`)
+  },
+
+  getCommandStatus: async (commandId: string) => {
+    const response = await apiClient.get<CommandJobStatusResponse>(
+      `/commands/jobs/${commandId}`
+    )
+    return response.data
+  },
+
+  /**
+   * Poll command status until completed or failed.
+   * Returns true if completed successfully, false if failed.
+   */
+  waitForCommand: async (
+    commandId: string,
+    options?: { maxAttempts?: number; intervalMs?: number }
+  ): Promise<boolean> => {
+    const maxAttempts = options?.maxAttempts ?? 60 // Default 60 attempts
+    const intervalMs = options?.intervalMs ?? 2000 // Default 2 seconds
+
+    for (let i = 0; i < maxAttempts; i++) {
+      try {
+        const status = await insightsApi.getCommandStatus(commandId)
+        if (status.status === 'completed') {
+          return true
+        }
+        if (status.status === 'failed' || status.status === 'canceled') {
+          console.error('Command failed:', status.error_message)
+          return false
+        }
+        // Still running, wait and retry
+        await new Promise(resolve => setTimeout(resolve, intervalMs))
+      } catch (error) {
+        console.error('Error checking command status:', error)
+        // Continue polling on error
+        await new Promise(resolve => setTimeout(resolve, intervalMs))
+      }
+    }
+    // Timeout
+    console.warn('Command polling timed out')
+    return false
   }
 }
\ No newline at end of file
diff --git a/frontend/src/lib/locales/en-US/index.ts b/frontend/src/lib/locales/en-US/index.ts
index 477ea3a..ca0bf49 100644
--- a/frontend/src/lib/locales/en-US/index.ts
+++ b/frontend/src/lib/locales/en-US/index.ts
@@ -357,6 +357,7 @@ export const enUS = {
     viewInsight: "View Insight",
     deleteInsight: "Delete Insight",
     deleteInsightConfirm: "Are you sure you want to delete this insight? This action cannot be undone.",
+    insightGenerationStarted: "Insight generation started. It will appear shortly.",
     deleteNoteConfirm: 'Are you sure you want to delete this note? This action cannot be undone.',
     editNote: 'Edit note',
     createNote: 'Create note',
diff --git a/frontend/src/lib/locales/ja-JP/index.ts b/frontend/src/lib/locales/ja-JP/index.ts
index 375aae1..49e13ea 100644
--- a/frontend/src/lib/locales/ja-JP/index.ts
+++ b/frontend/src/lib/locales/ja-JP/index.ts
@@ -357,6 +357,7 @@ export const jaJP = {
     viewInsight: "インサイトを表示",
     deleteInsight: "インサイトを削除",
     deleteInsightConfirm: "このインサイトを削除しますか？この操作は元に戻せません。",
+    insightGenerationStarted: "インサイトの生成が開始されました。まもなく表示されます。",
     deleteNoteConfirm: 'このノートを削除しますか？この操作は元に戻せません。',
     editNote: 'ノートを編集',
     createNote: 'ノートを作成',
diff --git a/frontend/src/lib/locales/pt-BR/index.ts b/frontend/src/lib/locales/pt-BR/index.ts
index ccffae8..803581d 100644
--- a/frontend/src/lib/locales/pt-BR/index.ts
+++ b/frontend/src/lib/locales/pt-BR/index.ts
@@ -357,6 +357,7 @@ export const ptBR = {
     viewInsight: "Ver Insight",
     deleteInsight: "Excluir Insight",
     deleteInsightConfirm: "Tem certeza que deseja excluir este insight? Esta ação não pode ser desfeita.",
+    insightGenerationStarted: "Geração de insight iniciada. Aparecerá em breve.",
     deleteNoteConfirm: "Tem certeza que deseja excluir esta nota? Esta ação não pode ser desfeita.",
     editNote: "Editar nota",
     createNote: "Criar nota",
diff --git a/frontend/src/lib/locales/zh-CN/index.ts b/frontend/src/lib/locales/zh-CN/index.ts
index b8397c6..23979e9 100644
--- a/frontend/src/lib/locales/zh-CN/index.ts
+++ b/frontend/src/lib/locales/zh-CN/index.ts
@@ -363,6 +363,7 @@ export const zhCN = {
     noNotebooksAvailable: "暂无可用笔记本",
     deleteInsight: "删除见解",
     deleteInsightConfirm: "确定要删除此见解吗？此操作无法撤销。",
+    insightGenerationStarted: "见解生成已开始，稍后将显示。",
     notEmbeddedAlert: "内容未嵌入向量",
     notEmbeddedDesc: "此内容尚未为了向量搜索进行嵌入。嵌入可以启用高级搜索功能并更好地发现内容。",
     openOnYoutube: "在 YouTube 上打开",
diff --git a/frontend/src/lib/locales/zh-TW/index.ts b/frontend/src/lib/locales/zh-TW/index.ts
index fa6c5a9..9346ee4 100644
--- a/frontend/src/lib/locales/zh-TW/index.ts
+++ b/frontend/src/lib/locales/zh-TW/index.ts
@@ -363,6 +363,7 @@ export const zhTW = {
     noNotebooksAvailable: "暫無可用筆記本",
     deleteInsight: "刪除見解",
     deleteInsightConfirm: "確定要刪除此見解嗎？此操作無法撤銷。",
+    insightGenerationStarted: "見解生成已開始，稍後將顯示。",
     notEmbeddedAlert: "內容未嵌入向量",
     notEmbeddedDesc: "此內容尚未為了向量搜尋進行嵌入。嵌入可以啟用進階搜尋功能並更好地發現內容。",
     openOnYoutube: "在 YouTube 上開啟",
diff --git a/open_notebook/domain/CLAUDE.md b/open_notebook/domain/CLAUDE.md
index b1b11cc..3ec6289 100644
--- a/open_notebook/domain/CLAUDE.md
+++ b/open_notebook/domain/CLAUDE.md
@@ -32,7 +32,7 @@ Two base classes support different persistence patterns: **ObjectModel** (mutabl
   - `vectorize()`: Submit async embedding job (returns command_id, fire-and-forget)
   - `get_status()`, `get_processing_progress()`: Track job via surreal_commands
   - `get_context()`: Returns summary for LLM context
-  - `add_insight()`: Generate and store insights with embeddings
+  - `add_insight()`: Submit async insight creation via `create_insight_command` (fire-and-forget, returns command_id)
 
 - **Note**: Standalone or linked notes
   - `save()`: Submits `embed_note` command after save (fire-and-forget)
@@ -80,7 +80,7 @@ Two base classes support different persistence patterns: **ObjectModel** (mutabl
 - **Auto-embedding behavior**:
   - `Note.save()` → auto-submits `embed_note` command
   - `Source.save()` → does NOT auto-submit (must call `vectorize()` explicitly)
-  - `Source.add_insight()` → auto-submits `embed_insight` command
+  - `Source.add_insight()` → submits `create_insight_command` which handles DB insert + `embed_insight` command (all fire-and-forget)
 - **Relationship strings**: Must match SurrealDB schema (reference, artifact, refers_to)
 
 ## How to Add New Model
diff --git a/open_notebook/domain/notebook.py b/open_notebook/domain/notebook.py
index f24802a..05dc8f2 100644
--- a/open_notebook/domain/notebook.py
+++ b/open_notebook/domain/notebook.py
@@ -454,53 +454,52 @@ class Source(ObjectModel):
             logger.exception(e)
             raise DatabaseOperationError(e)
 
-    async def add_insight(self, insight_type: str, content: str) -> Any:
+    async def add_insight(self, insight_type: str, content: str) -> Optional[str]:
         """
-        Add an insight to this source.
+        Submit insight creation as an async command (fire-and-forget).
 
-        Creates the insight record without embedding, then submits an async
-        embed_insight command to generate the embedding in the background.
+        Submits a create_insight command that handles database operations with
+        automatic retry logic for transaction conflicts. The command also submits
+        an embed_insight command for async embedding.
+
+        This method returns immediately after submitting the command - it does NOT
+        wait for the insight to be created. Use this for batch operations where
+        throughput is more important than immediate confirmation.
 
         Args:
             insight_type: Type/category of the insight
             content: The insight content text
 
         Returns:
-            The created insight record(s)
+            command_id for optional tracking, or None if submission failed
+
+        Raises:
+            InvalidInputError: If insight_type or content is empty
         """
         if not insight_type or not content:
             raise InvalidInputError("Insight type and content must be provided")
+
         try:
-            # Create insight WITHOUT embedding (fire-and-forget embedding via command)
-            result = await repo_query(
-                """
-                CREATE source_insight CONTENT {
-                        "source": $source_id,
-                        "insight_type": $insight_type,
-                        "content": $content,
-                };""",
+            # Submit create_insight command (fire-and-forget)
+            # Command handles retries internally for transaction conflicts
+            command_id = submit_command(
+                "open_notebook",
+                "create_insight",
                 {
-                    "source_id": ensure_record_id(self.id),
+                    "source_id": str(self.id),
                     "insight_type": insight_type,
                     "content": content,
                 },
             )
+            logger.info(
+                f"Submitted create_insight command {command_id} for source {self.id} "
+                f"(type={insight_type})"
+            )
+            return str(command_id)
 
-            # Submit embedding command (fire-and-forget)
-            if result and len(result) > 0:
-                insight_id = str(result[0].get("id", ""))
-                if insight_id:
-                    submit_command(
-                        "open_notebook",
-                        "embed_insight",
-                        {"insight_id": insight_id},
-                    )
-                    logger.debug(f"Submitted embed_insight command for {insight_id}")
-
-            return result
         except Exception as e:
-            logger.error(f"Error adding insight to source {self.id}: {str(e)}")
-            raise
+            logger.error(f"Error submitting create_insight for source {self.id}: {e}")
+            return None
 
     def _prepare_save_data(self) -> dict:
         """Override to ensure command field is always RecordID format for database"""