I'm trying to create a 'simple' agent that grabs a set of records from Airtable and creates a table within Google Docs. So basically table-to-table.
I have the following tools available for the agent:
- GetDocStructure
- InsertTableRow
As I understand it one would need to get the relevant doc in order to understand its structure and then append a row to the table within the doc. If I took the time to run through all the indexes within the document I can do it manually - by hard-coding variable to variable.
How come the smartest LLMs in the room that can vibe-code 3D games can't get this right?? Are my expectations to high or is my approach to low (key)?