AI native · interactive

Call a tool, then answer

The model does not have live weather. With a tool, it asks your app. Your app runs the tool and hands the result back. Without a tool, it can only guess.

User

Model

has 1 tool

Your app

runs the tool

Weather API

live data

Ask the model and watch where the request goes.

Step 1 of 5 · read left to right

1
Read the question
Some questions can be answered from the prompt. This one cannot.
What is the weather in Lahore right now?
2
Decide whether a tool is needed
The model picks a tool only if you gave it one. Otherwise it has to improvise.
get_weather can answer a city and a time of “now”.
3
Write the tool call
This is data for your app, not the answer the user sees.
get_weather({ "city": "Lahore" })
4
Run the tool in your app
The model does not fetch the weather. Your server does, then returns JSON.
{ "temp": 34, "unit": "C", "note": "sample reading" }
5
Answer from what came back
With a result, the sentence quotes it. Without one, the number is made up.
It is 34°C in Lahore right now.
Grounded in the sample reading.

In production

Tools and technologies

Models emit structured call requests; your app runs the tool and returns results — same pattern everywhere.

Not tied to one language — use these names when you search docs or design a system.