Question

A developer is building a customer-support summarization app on Azure AI Foundry. The model sometimes returns summaries in prose paragraphs, but the downstream system requires a strict JSON object with fixed keys ('issue', 'resolution', 'sentiment'). The developer wants the most reliable way to consistently enforce this exact output structure without fine-tuning. Which prompt engineering approach should they apply?

Accepted Answer

Provide a system message specifying the required JSON schema plus a few-shot example showing an input and its exact JSON output, and enable structured/JSON output mode.

Answer

Increase the temperature parameter to 1.0 so the model explores more diverse formatting options.

Answer

Set the max_tokens parameter to a very low value to force the model to be concise.

Answer

Send only the raw user transcript with no system message and rely on the model's default behavior.