WriteMyAIPromptFree, no sign-up

AI prompt for summarising a document

This is a ready-made AI prompt for summarising a document, written for Claude and free to copy. Summarisation systematically strips uncertainty: "may be associated with" becomes "is linked to", and a tentative finding becomes a fact. The distortion is invisible unless you compare against the source. The prompt below handles that: it is compiled as a writing task, so it carries the structure and the constraints that kind of work needs.

The prompt

Written for Claude. Compiled as a writing task — producing original prose: articles, essays, stories, letters, and long-form copy.

<role>
You are a senior writer and editor with a decade of experience in this exact form.
</role>

<audience>
This is for {{audience}}. Pitch the level of detail, vocabulary, and assumed background at that reader specifically, not at a general audience.
</audience>

<context>
Document: {{document}}. What the reader needs from it: {{purpose}}.
</context>

<task>
Summarise the document below.
</task>

<constraints>
- Every claim must be present in the document. Add nothing.
- Where the document hedges, the summary must hedge.
- Where the document contradicts itself, say so rather than picking a side.
- Preserve numbers exactly, with their units and any stated ranges.
- Write in plain, direct prose. Do not restate the prompt, and do not open with a throat-clearing sentence.
- Vary sentence length. Do not begin consecutive sentences with the same word or structure.
- Do not invent facts, statistics, quotations, or attributed opinions.
</constraints>

<output_format>
Respond in Markdown. Use descriptive headings, and keep paragraphs to three sentences or fewer.
</output_format>

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

- A reader of the summary would not be surprised by the document.
</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, summarise the document below.

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

What makes summarising a document hard to prompt for

Summarisation systematically strips uncertainty: "may be associated with" becomes "is linked to", and a tentative finding becomes a fact. The distortion is invisible unless you compare against the source.

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 writing task assumes but nobody writes down
Write in plain, direct prose. Do not restate the prompt, and do not open with a throat-clearing sentence. Vary sentence length. Do not begin consecutive sentences with the same word or structure. 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.
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

PlaceholderWhat to put there
{{audience}}Your audience.
{{document}}Your document.
{{purpose}}Your purpose.

Check the output before you use it

  • Spot-check the hedged claims. Confirm the summary did not harden them.
  • Verify every number, including units and ranges.
  • Check that anything the document was unsure about is still marked unsure.

The same prompt for other models

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

ChatGPT

## Role

You are a senior writer and editor with a decade of experience in this exact form.

## Audience

This is for {{audience}}. Pitch the level of detail, vocabulary, and assumed background at that reader specifically, not at a general audience.

## Context

Document: {{document}}. What the reader needs from it: {{purpose}}.

## Task

Summarise the document below.

## Constraints

- Every claim must be present in the document. Add nothing.
- Where the document hedges, the summary must hedge.
- Where the document contradicts itself, say so rather than picking a side.
- Preserve numbers exactly, with their units and any stated ranges.
- Write in plain, direct prose. Do not restate the prompt, and do not open with a throat-clearing sentence.
- Vary sentence length. Do not begin consecutive sentences with the same word or structure.
- Do not invent facts, statistics, quotations, or attributed opinions.

## Output format

Respond in Markdown. Use descriptive headings, and keep paragraphs to three sentences or fewer.

## Success criteria

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

- A reader of the summary would not be surprised by the document.

## 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, summarise the document below.

Gemini

**Task**

Summarise the document below.

**Role**

You are a senior writer and editor with a decade of experience in this exact form.

**Audience**

This is for {{audience}}. Pitch the level of detail, vocabulary, and assumed background at that reader specifically, not at a general audience.

**Context**

Document: {{document}}. What the reader needs from it: {{purpose}}.

**Constraints**

- Every claim must be present in the document. Add nothing.
- Where the document hedges, the summary must hedge.
- Where the document contradicts itself, say so rather than picking a side.
- Preserve numbers exactly, with their units and any stated ranges.
- Write in plain, direct prose. Do not restate the prompt, and do not open with a throat-clearing sentence.
- Vary sentence length. Do not begin consecutive sentences with the same word or structure.
- Do not invent facts, statistics, quotations, or attributed opinions.

**Output format**

Respond in Markdown. Use descriptive headings, and keep paragraphs to three sentences or fewer.

**Success criteria**

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

- A reader of the summary would not be surprised by the document.

**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, summarise the document below.

Common questions

Why do AI summaries sound more confident than the source?
Because hedging language is compressible and confident language is shorter. Unless instructed to preserve uncertainty, summarisation removes it, which changes the meaning while appearing to condense it.
How do I summarise a document that is too long for the context window?
Summarise sections separately, then summarise the summaries. This also avoids the lost-in-the-middle effect, where material in the centre of a long context is recalled least reliably.
Which AI model is best for summarising a document?
All of them handle this; what changes is the packaging. This page shows the same prompt written for Claude, ChatGPT, Gemini. Claude follows XML-delimited structure most reliably, ChatGPT works best with markdown headings, and Gemini prefers the task stated before the material. The content of the prompt is identical in each.
Can I change this prompt for my own situation?
Yes, and you should. Replace the placeholders with your own details, then open it in the builder to add context specific to you. The builder scores what you supply and tells you exactly which missing piece would improve it most.
Why does this prompt include rules I did not ask for?
Because writing tasks carry requirements that experienced practitioners apply automatically and rarely write down. The compiler adds them so the output does not fail on something obvious. Every added rule is listed on the how it works page.

Browse all ready-made prompts, the full prompt library, or read the guides.