LLM Structured Output Demo

By Stefan Kuhn: GitHub.

Turn unstructured text into JSON and a table (minimal example).

This page calls an LLM and requests structured output (JSON).
The response is parsed and rendered as a table.

Explanation

The prompts instruct the LLM to return only valid JSON with a stable schema.
This keeps the flow predictable: prompt → LLM JSON → parse → table.

Update the endpoint, API key, or model before sending the request. Nothing is stored.

LLM Provider URL API Key Model

Paste any customer message, then run the structured-output workflow.

Customer Message

These prompts demand structured JSON so the demo can parse it deterministically.

System Prompt User Prompt

The raw model response (pretty-printed when valid JSON).

Model Response

When parsing succeeds, the structured fields appear in this table.