4 step workflow
Analyze a long PDF and generate a summary
Extract structure, claims, and action items from long documents.
Workflow progress
Check off steps as you finish them. Progress stays in this browser.
Ask for structure first
Why use it
Long-context assistants can map the document before summarizing.
Prompt
Before summarizing, list the document structure, key sections, tables, and possible limitations.
Input example
A long PDF
Output example
Structure map and summary plan
What to watch
Check whether appendices, footnotes, and scanned pages were actually read.
Extract claims, evidence, and tables
Why use it
Separating claims from evidence prevents a fluent summary from hiding weak support.
Prompt
List the main claims, evidence, table references, assumptions, and limitations. Include page numbers when available.
Input example
Document structure and the PDF
Output example
A claim-evidence table with page references.
What to watch
Verify numerical values directly in the original tables.
Draft a layered summary
Why use it
A short overview plus detailed findings serves different reading needs.
Prompt
Write a three-layer summary: five-line overview, key findings with page references, and open questions. Use only the extracted evidence.
Input example
Verified claim-evidence table
Output example
A layered summary with traceable findings.
What to watch
Keep uncertainty and author limitations visible.
Step 4
Verify and create action items
Why use it
A final review separates what the document says from what the reader should do next.
Prompt
Audit the summary against the document and then list action items separately, each with owner, evidence, and unresolved question.
Input example
PDF and layered summary
Output example
Corrected summary and separate action list.
What to watch
Do not upload confidential PDFs to services not approved by the document owner.