Hi, i saw an exam question and i need help for this one:
In an AI Agent flow, in order to complete a task that requires executing two distinct tools (Actions) before providing the answer, how many LLM calls are required, at minimum, to process the intentions and generate the final conclusion?
A. 1 B. 2 C. 3 D. 4 And another one it called 2 times the same action (GetClinic) and one time another action ( GetClinicByInsurance) . This is to return how many clinics have that insurance.A. 1 B. 2 C. 3 D. 4
Thank you
In an AI Agent Flow, only 1 LLM call is required to execute multiple tools and generate a final conclusion, even when the flow involves multiple actions such as calling various tools or server actions sequentially. The AI dynamically reasons, uses the tools provided, and generates a final structured answer within that single LLM execution AI agent actions.
What about the second scenario: it called 2 times the same action (GetClinic) and one time another action ( GetClinicByInsurance) . This is to return how many clinics have that insurance.A. 1 B. 2 C. 3 D. 4
1 LLM call is required.
-Saicharan
1. A-1
2. D-4 (2 callGetClinic + 1 call GetClinicByInsurance + 1 call for final)
Hi, This question is pretty similar to this one. And unfortunately, the Bot is giving the wrong answer here and in the other question.I tried to explain why I think it's wrong in more detail in the other question, but here is a small summary.On both questions here, due to lack of context, it's not possible to define with absolute precision what will be the agent flow. But the MINIMUM number of calls to the Model (LLM), in both situations, is two (2).The agent call the model (first call). The model decides it need to call the actions (tools). There is no context to tell if it calls one first or if it decides that it need to call all of them in sequence. But it is possible that it returns an answer saying all of them need to be called in sequence.The Agent will then call the actions and after getting the results will call the Model (LLM) again (second call) passing the previous answer and the results. The Model will decide if it has the solution/final answer or if it needs another turn. So, at MINIMUM, not knowing how the actions are called, 2 requests for the Model are made, whenever the model requires the execution of actions (tools). There are only two scenarios where a a single call will be made:a) There are NO actions available to the Model.b) The model decide it does not need to call any action.If those are questions that came from the certification (shame on you :P lol), and has no more details, I think 2 is the correct answer. But if the question has more details that allows you to understand how the actions are called: all in sequence from the answer, or one from one answer, than after looking to the action results it decides to call it again, and so on, then the answer may be different.But if THERE are calls to actions, it will NEVER be only one call to the model. At minimum 2.
I need to correct my answer. And ask sorry to NEO. You were right. :(It seems when the agent returns the results of the actions to the LLM, it is not calling it again, just returning what the LLM asked. So it does not count towards the number of calls to the LLM.This is a distinction I didn't understood initially.Not knowing if the actions were called all at once or in different steps of the reason-act loop means that in theory, as they can be called all at once, the minimum number of calls in both scenarios is, indeed 1.