diff --git a/assets/image_1769673122545_0.png b/assets/image_1769673122545_0.png
new file mode 100644
index 00000000..993f6b07
Binary files /dev/null and b/assets/image_1769673122545_0.png differ
diff --git a/journals/2026_01_29.md b/journals/2026_01_29.md
index 1909f2ec..a6c3f5b5 100644
--- a/journals/2026_01_29.md
+++ b/journals/2026_01_29.md
@@ -31,7 +31,9 @@ boox-notes:: ![BOOX Notes](../assets/../../onyx/TabUltraCPro/Notepad/JOURNALING.
 			- ![image.png](../assets/image_1769672971263_0.png)
 			- ![image.png](../assets/image_1769673005517_0.png)
 				- You said that the datasets are necessary to train the model. But it is sure that the existing models are nothing to do with agents or multi-agent systems. Isn't it? [[question]]
-			-
+			- ![image.png](../assets/image_1769673122545_0.png)
+				- I think you should focus on the workflow benchmarks [[question]]
+		-
 			- ### Questions
 				- {{query (and [[question]] [[29-01-2026]] )[[question]]}}
 				  query-table:: true
\ No newline at end of file