Which tool-call pattern reduces agent latency when calls do not depend on each other?
Why C is the answer
Running the calls in parallel
Independent calls need not wait for each other.
In detail
- Calls that do not depend on each other can run together.
- Dependent calls must stay in order.
Use case
An illustrative scenario, not a real incident.
An agent checks weather and flights at once, because neither depends on the other.
See it in different ways
sequenceDiagram participant U as Users participant L as Rate and budget limits participant A as Agent replicas participant S as External state participant M as GPU inference U->>L: Requests L->>A: Within limits A->>S: Reads and saves state A->>M: Model calls, in parallel if independent
flowchart TD
Q{"Does one call need the other's result?"}
Q -->|"Yes"| A0["Run in order"]
Q -->|"No"| A1["Run in parallel"]Key point: Parallel calls reduce latency.
Why the other options are wrong
- A. Asking the user to confirm each read
- That adds waiting.
- B. Repeating each call to compare results
- That adds cost.
- D. Running the calls one after another
- That is what makes it slow.