Unexpected LLM Call in “Null Agent” Workflow

[bkermen]

I created a “Null Agent” that only contains a Start and an End block and does absolutely nothing.

However, I’ve noticed in the Usage Log that every time this workflow runs, there’s a Claude 3.5 Haiku LLM call recorded, along with a small associated cost.

I’m certain the workflow doesn’t use any LLM model, and I even switched the underlying workflow model to GPT-5, suspecting that might be the cause - but this behavior persists.

Could someone clarify why this happens? The cost is negligible for normal use, but it could become significant if the workflow is scheduled to run frequently for a long time (e.g., every minute for a year).

[Alex_MindStudio]

Hi @bkermen,

That LLM call is used to generate the Run Title for each run. You can change the model used for that by updating the default Model in your Workspace settings:
https://app.mindstudio.ai/workspace/settings