Summarising Massive Documents

Your boss just sent you a 47-page government report and wants your thoughts by 3pm.

Old way: 2 hours of painful skimming, hoping you didn't miss the important part on page 31.

New way: paste it into Claude or ChatGPT, ask the right question, and get a 5-point briefing in 20 seconds.

IMAGE PLACEHOLDER
DALL-E 3 / Midjourney / Ideogram
A dramatic split-screen illustration: on the left, a towering stack of dense printed documents with a tiny overwhelmed person at the bottom. On the right, a single clean one-page summary with clear bullet points glowing on a dark background. Minimalist flat design, teal accent colours. No text.

The Document Summarisation Toolkit

The Quick Brief

"Summarise this document in 5 bullet points. Focus on the key findings, recommendations, and any action items. Each bullet should be under 2 sentences."

The Executive Summary

"You are a senior analyst. Read this report and write a 200-word executive summary for a busy CEO. Lead with the most important finding. Use plain language, no jargon."

The Specific Question

"Based on this document, answer the following question: [your specific question]. Quote the relevant section if possible."

The Action Extractor

"List every action item, deadline, or recommendation mentioned in this document. Format as a table with columns: Action | Owner (if mentioned) | Deadline (if mentioned)."

The Contradiction Finder

"Read this report and identify any claims or data points that seem inconsistent or contradictory. List them with the relevant page or section."

For Different Document Types

The "Talk to Your Document" Technique

Once you've pasted a document, you can ask follow-up questions like it's a conversation:

1. Paste the document: "Here is a document I need help with: [paste]"
2. Ask your first question: "What is the main argument?"
3. Then drill down: "What evidence do they give for that claim?"
4. Then challenge it: "What counterarguments are not addressed?"
5. Then apply it: "What should I do differently based on this?"

Interactive Exercise: Text Processing

This exercise shows you how to find the most common words in a piece of text — useful for quickly understanding what a document is about:

python-runner.pyInteractive
text = """
Artificial intelligence is transforming how businesses operate.
Companies that invest in AI tools see significant productivity gains.
However, AI implementation requires careful planning and change management.
The most successful AI deployments combine technology with strong human oversight.
Productivity and planning are the two most cited factors in successful AI adoption.
"""

from collections import Counter
import re

# Simple word frequency (excluding common words)
stop_words = {"the", "a", "an", "is", "in", "and", "with", "of", "that", "are", "for", "how", "see", "most"}
words = re.findall(r'\b[a-z]{4,}\b', text.lower())
filtered = [w for w in words if w not in stop_words]
top_words = Counter(filtered).most_common(5)

print("Top 5 keywords in this document:")
for word, count in top_words:
    print(f"  {word}: {count} mentions")
 Loading Kernel...

The "Too Long; Didn't Read" Chain

For very long documents (books, lengthy reports), use this chain:

1. "Summarise this document in 10 bullet points" (get the big picture)
2. "Which 2 of those points are most important for someone in [your role]?" (filter by relevance)
3. "Expand on point [X] with more detail" (drill into what matters)
4. "What should I do based on this?" (turn insight into action)

Each step builds on the last. You go from 50 pages to a personal action plan in under 5 minutes.

Key takeaway: Documents are no longer obstacles — they're inputs. Give AI the context, ask the right questions, and extract exactly what you need.

Up next: Lesson 7 — Basic data crunching without a single line of code