Webhook response returns full OpenAI JSON instead of plain text (OpenAI “Generate a response” module)

Questions
,
1

:bullseye: What is your goal?

To identify the exact, supported way in Make to return ONLY the assistant’s plain-text reply from the OpenAI “Generate a response” module via a Webhook response — with no JSON, no metadata, no arrays.

:thinking: What is the problem & what have you tried?

Webhook response returns full OpenAI JSON instead of plain text (OpenAI “Generate a response” module)

Hello Make Support,

I’m building a simple scenario and I can’t get a plain-text response back from a webhook. Even when I map the OpenAI module’s “Result” field, the webhook response still returns the full structured JSON payload.

Scenario (3 modules):

Webhooks → Custom webhook

OpenAI (ChatGPT, Sora, DALL-E, Whisper) → Generate a response

Model: o4-mini (system)

Prompt type: Text prompt

Webhooks → Webhook response (status 200)

Test:
I call the webhook in a browser: …hook…?text=hello

Expected:
Webhook response returns only the assistant message as plain text.

Actual:
Webhook response returns a JSON array/object including id/type/content/role, etc. (structured output). Mapping “Result” still returns structured data, not plain text.

Question:

What is the correct mapping path/field to return only the assistant message text from this OpenAI module?

Is “Result” supposed to be plain text? If yes, why is it returning structured JSON?

If this module does not expose message text directly, what is the recommended Make-native approach (e.g., JSON Parse, different OpenAI module, settings) to extract only the text?

Thanks.