Summary

Martin Fowler describes a pattern he calls 'Interrogatory LLM,' where instead of writing extensive context documents for LLMs, you prompt the LLM to interview you and extract the needed information through conversation. He credits Harper Reed for popularizing this approach, noting the importance of having the LLM ask one question at a time. Fowler extends the pattern to document review, where an LLM interviews domain experts to validate specifications rather than asking them to read and critique documents directly. He argues this technique has broad applicability beyond LLM workflows, serving as a writing aid for people who struggle to get knowledge out of their heads into written form.

Key Insight

Rather than laboriously writing context for LLMs or documents for humans, you can flip the interaction and have the LLM interview you, turning conversation into structured knowledge.

Spicy Quotes (click to share)

• 3

  The obvious way to do this is for a human to write this context, but an alternative is to use an LLM to write this context after interviewing a human.

• 4

  People often find reviewing hard, so a conversation with an LLM might be more fruitful, particularly if the document isn't well-written.

• 2

  To really understand something, I need to write about it. But different people are different. Many folks find writing hard, often very hard.

• 2

  This can be a real problem when we need to get information out of someone's head into a form that other humans can consume.

• 3

  Maybe such people would find it easier to ask an LLM to interview them than to write a document themselves.

• 5

  Certainly the result will have that tang of AI-writing that folks like me shudder at - but that's better than not having the information itself, either due to rushed writing or no writing at all.

Tone

reflective