Resource
Executive Summary Prompt
Summary, Why It Matters, What Happens Next — the executive summary contract for readers who will never open the source.
Stop AI Summaries from Making Things Up
The fidelity ladder: Balanced, High, Strict — and the six-rule battery that keeps a summary inside its source.
Overview
Summary hallucination is quiet: a connective sentence here, a "presumably" there, a gap filled with plausible context — and the summary now says things the source never did. The fix is a fidelity contract, and this resource demonstrates the strictest rung on the most ordinary source: weekly status updates. Strict Fidelity's battery — do not invent, do not infer unsupported conclusions, do not add missing context, only summarize what appears, preserve numbers and names exactly, and write "Not covered in the source" for empty sections — turns even a mundane roundup into a document you can trust without re-reading the source.
How to use this resource
-
Match the rung to the stakes
Balanced for internal convenience, High when numbers and names matter, Strict when the summary IS the record.
-
Swap fidelity and diff the prompt
Load this setup, change the fidelity level, and watch the FIDELITY RULES block rewrite itself — three rungs, one axis.
-
Keep the gap rule
"Not covered in the source" is the cheapest hallucination insurance in the whole prompt — never delete it.
Why This Works
- Naming the failure ("do not add missing context") works better than asking for accuracy
- The gap rule converts invention pressure into honest blanks
- Demonstrating strict mode on a mundane source proves fidelity is about trust, not document type
Best for
- Summaries that get forwarded without the source attached
- Teams who've caught a summary saying more than the source
- Mundane sources too — status updates get misquoted as much as contracts
Not for
- Creative rewriting or expansion — fidelity rules exist to prevent exactly that
- Fact-checking the source itself — fidelity keeps the summary inside the source, right or wrong
Use cases
- Hardening any summary prompt against quiet invention
- Choosing the right fidelity rung for the source's stakes
- Making gaps visible instead of plausibly filled
FAQ
How is Strict Fidelity different from the Balanced and High rungs in this ladder?
The Balanced and High rungs still lean on trust; Strict replaces the FIDELITY RULES block with the full six-rule battery — do not invent, do not infer unsupported conclusions, do not add missing context, only summarize what appears, preserve numbers/dates/names exactly, and write 'Not covered in the source' for empty sections. The other rungs relax one or more of those six; Strict enforces all of them at once.
Why does the summary say 'Not covered in the source' instead of just leaving a section blank?
That literal string is the gap rule doing its job: instead of padding a thin ## Key Points or inventing context to fill space, the prompt tells the model to write 'Not covered in the source' for any section with nothing behind it. The result is a visible blank rather than a plausible fill, so a forwarded summary never quietly implies coverage the source never had.
Can this fidelity prompt catch a wrong number or a false claim in the source itself?
It only keeps the summary inside the source, not the source inside reality. The rule 'preserve numbers, dates, and names exactly as written' copies a mistaken figure faithfully rather than correcting it, and no rule fact-checks the original. Treat fidelity as anti-hallucination for the summary layer; verifying the source's own accuracy stays a separate step you own wherever you run the prompt.
More resources from Structured Summary Prompt
Resource
Structured Summary Prompt — the Anatomy
The blocks a reliable summary prompt needs: source guidance, a fixed section skeleton, length budgets, fidelity rules, and quote handling.
Resource
Customer Call Summary Prompt
Action items only: every commitment from the call — owner, task, deadline — straight into the CRM, with quotes for the promises that matter.
Resources that pair well
Resource
Markdown Output Prompt — Lock the Document Structure
The contract that stops AI documents from restructuring themselves: a pinned section skeleton, forced tables, and strict consistency rules.
Resource
Extract Data From Text with AI
Free text in, named fields out. The extraction prompt pattern that turns any unstructured text into consistent, parseable records.
Resource
Fix a Vague Prompt
'Make it good', 'be detailed', 'keep it interesting' — vague prompts get vague output. The fix is mechanical: replace every fuzzy word with a checkable instruction.
Related tools
Tool
Structured Summary Prompt
Build summary prompts with fixed sections, length caps, and no-invention fidelity rules.
Guides for this resource
Guide
Summarize a Document With AI Without Distorting It
AI summaries read clean but quietly distort — an exception dropped, a "may" turned to "will," a target read as a guarantee. Here's how to summarize a document faithfully: scope it, name what must be preserved, keep the caveats, and check it back against the text.
Guide
Build an Evidence Table From Sources With AI
Ask AI to "summarize these sources and list the findings" and one interview becomes "customers," a real disagreement disappears, and no claim shows its source. Build an evidence table instead: one row per claim, with its source, exact quote, and how strong the support really is.