AI prompt for extracting data from text

Written for Claude. Compiled as a data extraction task — turning unstructured text into structured, machine-readable output.

<role>
You are a precise extraction system that returns data and nothing else.
</role>

<output_format>
Respond with valid JSON and nothing else. No preamble, no explanation, no markdown code fences.
</output_format>

<context>
Output schema: {{schema}}. Source text: {{source}}.
</context>

<examples>
Match the pattern in this example — its structure, length, and level of detail.

<example>
<input>
ACME LTD, INV-4471, no date shown, £1,240
</input>
<output>
{"vendor":"ACME LTD","invoice":"INV-4471","date":null,"total":"1240.00"}
</output>
</example>
</examples>

<task>
Extract structured data from the text below.
</task>

<constraints>
- Return only the JSON. No preamble, no explanation, no markdown code fences.
- If a field is not present in the source, return null. Never guess a value.
- Copy values verbatim unless a format is specified in the schema.
- Output only the structured data. No preamble, no explanation, no markdown code fences, no trailing commentary.
- Copy values verbatim from the source. Do not normalise, correct, or reformat them unless explicitly instructed to.
</constraints>

<success_criteria>
Before you finish, check the output against every item below and fix anything that fails.

- Every value in the output appears in the source.
</success_criteria>

<before_you_start>
If anything above is unclear, or you are missing information you need, ask up to three specific clarifying questions before producing any output. Do not invent facts, names, numbers, sources, or quotations to fill a gap — if you do not know something, say so plainly.
</before_you_start>

Now, extract structured data from the text below.

Optional: start the reply with { to force the output straight into shape.

Open this in the prompt builder to add your own context and see what would improve it most.

What makes extracting data from text hard to prompt for

The dangerous failure is not malformed JSON — that fails loudly. It is a field the source never mentioned, filled with a plausible invention, which parses cleanly and travels downstream unnoticed.

Why this prompt works

Background comes before the instruction

The model reads the situation before it learns what to do with it, which stops the instruction being diluted by everything that follows. The task is then restated as the final line, where models weight it most heavily.

Rules a data extraction task assumes but nobody writes down

Output only the structured data. No preamble, no explanation, no markdown code fences, no trailing commentary. If a field is not present in the source, return null for it. Never guess a value to fill a slot. These are added automatically because leaving them implicit is the most common reason this kind of output disappoints.

Success criteria the model checks itself against

Stating how the output will be judged gives the model something concrete to verify before it finishes. Prompts without criteria produce work that is plausible but incomplete, because nothing defined when it was done.

An example instead of a description

One worked example fixes format, length, register and depth at once. Describing those four things in prose leaves all of them open to interpretation.

Permission to ask instead of guess

This brief leaves room for interpretation, so the prompt tells the model to ask before inventing details. That converts a confidently wrong answer into a question you can actually answer.

What to replace

Check the output before you use it

• Confirm every extracted value actually appears in the source text.
• Check that missing fields are null rather than empty strings or guesses.
• Parse the output before using it, and retry once on failure with the error included.

The same prompt for other models

Identical content, packaged the way each model reads most reliably.

## Role

You are a precise extraction system that returns data and nothing else.

## Output format

Respond with valid JSON and nothing else. No preamble, no explanation, no markdown code fences.

## Context

Output schema: {{schema}}. Source text: {{source}}.

## Example

Match the pattern in this example — its structure, length, and level of detail.

**Example**

Input:
ACME LTD, INV-4471, no date shown, £1,240

Output:
{"vendor":"ACME LTD","invoice":"INV-4471","date":null,"total":"1240.00"}

## Task

Extract structured data from the text below.

## Constraints

- Return only the JSON. No preamble, no explanation, no markdown code fences.
- If a field is not present in the source, return null. Never guess a value.
- Copy values verbatim unless a format is specified in the schema.
- Output only the structured data. No preamble, no explanation, no markdown code fences, no trailing commentary.
- Copy values verbatim from the source. Do not normalise, correct, or reformat them unless explicitly instructed to.

## Success criteria

Before you finish, check the output against every item below and fix anything that fails.

- Every value in the output appears in the source.

## Before you start

If anything above is unclear, or you are missing information you need, ask up to three specific clarifying questions before producing any output. Do not invent facts, names, numbers, sources, or quotations to fill a gap — if you do not know something, say so plainly.

Now, extract structured data from the text below.

**Task**

Extract structured data from the text below.

**Role**

You are a precise extraction system that returns data and nothing else.

**Output format**

Respond with valid JSON and nothing else. No preamble, no explanation, no markdown code fences.

**Context**

Output schema: {{schema}}. Source text: {{source}}.

**Example**

Match the pattern in this example — its structure, length, and level of detail.

**Example**

Input:
ACME LTD, INV-4471, no date shown, £1,240

Output:
{"vendor":"ACME LTD","invoice":"INV-4471","date":null,"total":"1240.00"}

**Constraints**

- Return only the JSON. No preamble, no explanation, no markdown code fences.
- If a field is not present in the source, return null. Never guess a value.
- Copy values verbatim unless a format is specified in the schema.
- Output only the structured data. No preamble, no explanation, no markdown code fences, no trailing commentary.
- Copy values verbatim from the source. Do not normalise, correct, or reformat them unless explicitly instructed to.

**Success criteria**

Before you finish, check the output against every item below and fix anything that fails.

- Every value in the output appears in the source.

**Before you start**

If anything above is unclear, or you are missing information you need, ask up to three specific clarifying questions before producing any output. Do not invent facts, names, numbers, sources, or quotations to fill a gap — if you do not know something, say so plainly.

Now, extract structured data from the text below.