I think my mental model of what Tools
I think my mental model of what Tools are is all wrong. I have imagined them as utility functions the model can chose to use to enrich its prompt/help solve the problem by building knowledge step by step. But it seems it is much more like a gateway for a more concrete 1:1 question -> answer flow.
I was thinking that a prompt would have 2-3 tools listed and run them in a sequence to fill in blanks to solve the main big problem in the prompt. But that is probably too wishful thinking.
Is it more correct to say that from a list of x tools, the model can chose one to run with if it can't produce the "answer" to the prompt by itself? It will always try without first, and based on the result it might redirect the original prompt to a tool and use its result as the real result (and possibly fail with the first tool and if so continue down the list)
I was thinking that a prompt would have 2-3 tools listed and run them in a sequence to fill in blanks to solve the main big problem in the prompt. But that is probably too wishful thinking.
Is it more correct to say that from a list of x tools, the model can chose one to run with if it can't produce the "answer" to the prompt by itself? It will always try without first, and based on the result it might redirect the original prompt to a tool and use its result as the real result (and possibly fail with the first tool and if so continue down the list)
