TODAY'S EDITION
BrandBloom (Beurendeublum)
Instorm Technical Column
Why don’t websites costing millions of won show up in AI answers?
Because AI doesn’t collect text in the same way people see it. We’ve laid out the findings as reported in a Princeton University KDD 2024 paper, which tested “writing that AI cites” using 10,000 queries. Part 3 of the series.
AI Has Two Kinds of Memory—and Both Require “Accumulation”“I’ve published posts, so why doesn’t AI know about us yet?” The answer becomes clear when you distinguish the two ways AI gets information—learned memory (pretraining) and real-time search (RAG). We explain why accumulation takes time, based on official Google documentation and LLM training principles. Part 4 of the series.
Same Question, Different Answer—The Technical Reasons AI Answers Change Every TimeAsk ChatGPT the same question twice and you’ll get different answers. It’s not a bug; it’s by design. We explain with facts why “100% guaranteed AI visibility” is impossible, based on four variables: temperature, probabilistic sampling, search fluctuations, and model updates. Part 2 of the series.
Google said, “There’s no need for GEO tricks”—and we agree.In May 2026, Google published its official guide to optimizing for AI search. We mostly agree with its conclusion that “AEO and GEO are ultimately SEO.” However, we’ve used the original guide and data to outline three assumptions it doesn’t mention that don’t align with real-world practice in the Korean market. A special edition of our series.A guide to how AI gathers information and selects sources, and how to check whether your content is actually appearing.
A comparison of top-ranking Naver blogs and AI sources across 16 keywords. Why search rankings and citations should be considered separately.
Ask us anything about our service
Try asking