Hi everyone,
I’m hitting a roadblock with the Looker Conversation Agent (GenAI) and was wondering if anyone has encountered something similar.
The Problem: When I ask the agent to filter data for a specific Project ID (e.g., ai_Chat), the agent seems to automatically inject special escape characters or prefixes (like ^) into the value during query execution. This causes the dataset lookup to fail.
Error Message:
“Failed to locate the dataset for Project ai_Chat. Please verify the ID is correct and ensure no special escape characters (like ^) were injected.”
What I’ve Checked:
-
The underlying dimension
project_idis defined as atype: stringin LookML. -
In a standard Looker Explore, the filter works perfectly.
-
The issue only occurs when interacting through the Conversation Agent / AI interface.
What I Need Help With:
-
Is there a way to force the Agent to treat certain dimensions as plain text without any auto-escaping logic?
-
Is this a known behavior with how the agent handles underscores (
_) or specific string patterns?
