Deep Insight // Try thisAll worksheets  /  Chapter 6

Chapter 6: Steering the Ship | Toolkit

Prompt Lab: Vague to Mission Ready


Autosaves on this device PDF version

Bad prompts are still useful. They show you how the model behaves. This lab walks one prompt from vague to mission ready, one deliberate change at a time.

Part 1

What kind of prompt is this?

TypeThe question it asksExample
RetrievalWhat is already known?What are the known IOCs for Shiny Hunters?
CompressionWhat is important?Summarize this 50-page threat report into key findings
InterpretationWhat does this mean?What does the shift in Sandworm's targeting pattern in the last 30 days suggest about their priorities?
GenerationCan you create something new?Develop a scenario where Iran targets U.S. critical infrastructure during a geopolitical crisis
JudgmentWhat's the answer?Is Charming Kitten responsible for the infrastructure outage? (Always ask what supports and contradicts it.)
My task is mostly:
Part 2

Earn all four badges

Part 3

Iteration log

Start vague on purpose. Each round, change one thing and note what broke and what you fixed.

Version 1 (the vague one)
Version 2
Version 3
Version 4
Part 4

Advanced techniques used

Pro tip: if your prompt is terrible, ask the model to help fix it.

Deep Insight: AI for the Intelligence Analyst by Rae Bakerdeepinsightbook.com