Working with GPT

Last modified: 19 December 2024

Integration Dasha with OpenAI GPT

Dasha Scripting Language allows you to integrate GPT flow with common dialog flow using #answerWithGPT function.

How does it work?

When you want to response to the user with GPT, you can call

var result = #answerWithGPT(`Your prompt`, interruptible:true, gptOptions: {
            model:"openai/gpt-4",
            openai_apikey: "YOUR APIKEY"
        }, sayOptions: {
            interruptDelay: 1.0
        });

Where model is either:

openai/gpt-4
openai/gpt-4o-mini
openai/gpt-3.5-turbo
any other OpenAI model using prefix openai/

openai_apikey - your OpenAI APIKEY

interruptible & interruptDelay - allows user to interrupt an answer process, if user is speaking for interruptDelay number of seconds

the result will contain information about what happened:

result.interrupted will be true, if user has interrupted the system
result.functionCalled will be true, if gpt requested a function call (see docs below)

Asking GPT without reply to the customer

You can use #askGPT, which is simular to the #answerWithGPT for getting response from the GPT or for calling your functions.

Found a mistake? Let us know.

Working with GPT

Integration Dasha with OpenAI GPT

Asking GPT without reply to the customer

Enroll in beta