How to Use Parse & Pack

Parse & Pack helps you upload one or more PDFs, describe exactly what you want to keep, and get back one or more filtered PDFs.

The app uses AI to decide which pages match your criteria, and it can also group matching pages into category PDFs (for example, one PDF for “trigonometry” and one PDF for “calculus”).

Step-by-step

  • Sign in using the button in the top-right.
  • Upload your PDF using “Upload Files”. You can upload one or multiple PDFs.
  • Describe what to keep in the text box. Be specific (examples below).
  • Send to start processing. You’ll see progress while the document is extracted and analyzed.
  • Download the output(s) when it finishes. If multiple PDFs are generated, your browser may download multiple files.

What you get back (outputs)

  • If your criteria matches pages, Parse & Pack generates one output PDF per category.
  • If a category has zero matching pages, no PDF is created for that category.
  • If nothing matches at all, you’ll get a reasoning report PDF explaining why.
  • Page order inside each output is deterministic: it follows the AI's requested order when provided, otherwise it defaults to the original page order.

How to write prompts (the important part)

Think of your prompt as a set of rules. The clearer and more testable your rules are, the better the output will be.

  • State the goal: what the output is for (practice, revision, exam prep).
  • Specify inclusion rules: what to keep (topics, keywords, question types).
  • Specify exclusion rules: what to remove (solutions, answer keys, cover pages, ads).
  • Define categories if you want grouped PDFs (e.g. trig vs calculus vs geometry).
  • Be explicit about edge cases: diagrams, tables, “worked examples”, mixed-topic pages.

Example prompts

  • Single topic filter: “Keep only trigonometry questions. Exclude solutions and answer keys.”
  • Category buckets: “Categorize into two PDFs: Trigonometry and Calculus. Only keep question pages (no solutions). If a page contains both, put it in the category with more questions on that page.”
  • Question-type focus: “Keep multiple-choice questions only. Exclude short-answer, explanations, and marking guidelines.”
  • Diagram handling: “Keep pages with geometry questions and any page that contains a diagram referenced by a kept question.”
  • Ordering: “Within each category, order easier questions first, then harder.”

Using multiple documents

  • When you upload multiple PDFs, the app analyzes each file and then groups kept pages into your categories across all uploads.
  • If you ask for categories (like trig/calculus), you can upload 5 documents and still get back 2 output PDFs.
  • If you don't mention categories, the app may still place kept pages into “uncategorized”. If you want strict outputs, specify categories.

Example prompts

  • Homework: “Keep only pages that contain math problems. Remove answer keys.”
  • Study guide: “Keep definitions and theorems. Remove examples and ads.”
  • Exam prep: “Keep multiple-choice questions and any diagrams referenced.”
  • Worksheets: “Keep pages that have practice questions. Drop instructions pages.”

Tips

  • If your PDF is large, extraction may happen locally in your browser for performance.
  • Use clear keywords (e.g., “questions”, “exercises”, “diagrams”, “solutions”) to get more accurate filtering.
  • If the output isn't what you expected, tighten the rules: name the topic, list what to exclude, and give 1–2 examples of what counts.
  • If your browser blocks multiple downloads, allow downloads for this site (common when generating multiple category PDFs).

Common errors (and what they mean)

  • “Your upload exceeds your balance.” The total pages you uploaded is higher than your remaining page balance. Upload fewer pages or top up.
  • “Authentication required” You're not signed in (or your session expired). Sign in and retry.
  • “PDF is too large for server-side extraction.” The server refused extraction for a large file to stay within Worker limits. Usually the app will extract locally in your browser instead.
  • “Classification failed …” The AI step failed. This can happen due to network issues, provider timeouts, or temporary rate limits. Retrying often works.
  • “Extraction failed …” The PDF couldn't be read or parsed. Try re-exporting the PDF or using a different source file.

Back to Home.