Hi there!
We’ve got an exciting update for our MCP Product! 
In most of the industry, MCP tools only work for simple tasks like sending a Slack message or an email. 
But when it comes to more complex actions—like adding a new row in Google Sheets—they often fail. 
Why does this happen?
Because the usual prompt is something like:
“Insert this sales lead into my sales sheet.”
But the LLM doesn’t know:
What your Google Sheet looks like
What your column names are
What your sheet is called
And if you try using auto-generated MCP or similar automation tools, they usually break, because they are not designed carefully for LLM.
Our Solution
We’ve added a mini agent inside MCP actions. 
This agent first understands the goal, then explores your account to gather full context before taking any action. After that, it performs the execute the tool correctly!
It understands the context just like a human using the interface. It selects the right options, waits for new ones to load, and keeps going until everything is filled out.
Why does this matter?
It means MCP product become much stronger, which means Agents tools become way powerful at performing more complex tasks.
Note: The mini agents uses AI credits (gpto-mini) model to do necessary thinking, You can find them under AI usage table.
Happy Automating!
Hazem.