New 'Interrogatory LLM' Technique Lets AI Interview Humans for Complex Tasks

Breaking: AI Interviews Replace Manual Context Writing

A novel approach called 'interrogatory LLM' is revolutionizing how artificial intelligence handles complex tasks. Instead of requiring humans to write extensive context documents, the LLM interviews the human directly, asking targeted questions to gather all necessary information.

This technique, first described by engineer Harper Reed, flips the traditional workflow. The AI becomes the interrogator, probing for details about feature design, implementation guidelines, and external systems—allowing it to generate a comprehensive context report for subsequent model sessions.

'The LLM should ask me all the questions it needs to create this appropriate context,' said Martin Fowler, a prominent software thought leader who expanded on Reed's concept. 'I can feed much of the information it needs and tell it other sources to consult.'

Key Breakthrough: One Question at a Time

Reed's method insists on a critical constraint: the LLM must ask only one question per turn. Fowler noted that frequent reminders are needed to keep the model from overwhelming users with multiple queries. This focused approach ensures the human can provide clear, undivided answers.

The interrogatory LLM can also be used for document review. Given a software specification, the model interviews an expert to verify accuracy—a more engaging alternative to asking the expert to read and critique a potentially poorly written document.

Background

Complex tasks like designing new features require vast context: user descriptions, implementation guidelines, and external system data—often several pages of markdown. Traditionally, a human must write this context, a time-consuming process that many find difficult.

People who struggle with writing—whether due to time constraints or cognitive load—often leave knowledge unshared. The interrogatory LLM bridges this gap by letting the AI extract information conversationally, producing a usable document even if the human never writes a word.

Broader Implications for Knowledge Workers

Fowler noted that the technique extends beyond narrow LLM use cases. 'Many folks find writing hard, often very hard,' he said. 'Maybe such people would find it easier to ask an LLM to interview them than to write a document themselves.' While the resulting text may carry an 'AI-writing tang,' it is better than missing information entirely due to rushed or absent documentation.

What This Means

This development could dramatically lower the barrier to capturing expert knowledge. Organizations can now extract insights from non-writers without forcing them to produce polished prose. The same interview method can also streamline document creation and review cycles, reducing errors and saving time.

As the technique gains traction, expect wider adoption in software development, research, and any field where tacit knowledge needs to be formalized. The interrogatory LLM may become a standard tool for human-AI collaboration, making complex task execution more accessible and efficient.