EGI — Guided Interactive Exploration

Atomic Claim

Overview

EGI is a multi-turn AI interaction pattern that systematically maps an unfamiliar knowledge domain through structured questioning sequences, preventing both premature closure and infinite exploration.

💡 The Pattern

EGI is used when entering an unfamiliar topic. It proceeds in three moves:

Move 1 — Panoramic mapping

"Give me a landscape map of [topic area]. What are the 5-7 major 
sub-areas, the key open questions in each, and the 2-3 most 
cited researchers or papers per area? Format as a structured outline."

Move 2 — Targeted drilling

"In the sub-area of [X from Move 1], what are the 3 most important 
papers published since 2022? For each: main contribution, 
key metric, and whether it is reproducible."

Move 3 — Gap identification

"Based on what you've described, what are the most significant 
gaps in the literature on [X]? Which gaps are most tractable 
for a 6-month MSc thesis with local compute?"

🧩 Application to PUMA

Used in F0 (Initiation) to map: LLM benchmarks landscape, PM+AI literature, prompting strategies for classification tasks. See: 60 - Resources/61 Prompts/61.1 LLM-Tools/PT-Claude-EGI-Exploration

🔗 Connected Ideas

Part of: PN-MIT-Student-Method | Uses: PN-RCOIF-Framework | Followed by: PN-AMI-Framework