2ndvector

2ndvector is a personal research lab focused on a single practical question: when answers are generated by AI rather than returned as links, what determines who shows up — and how accurately?

The lab sits at the intersection of search, language models, and brand representation. Not the hype version of that intersection. The operational one.

What I’m Investigating

How AI systems retrieve and cite information

ChatGPT, Perplexity, Claude, Google AI Overviews — each surfaces information differently. I track how these systems select sources, construct citations, and synthesize answers, looking for the underlying patterns that determine visibility.

The gap between what a brand is and how AI describes it

AI systems often get things subtly wrong: outdated facts, mixed-up attributions, flattened nuance. Understanding why that gap exists — and what closes it — is one of the more interesting problems in search right now.

GEO and AEO as disciplines

Generative Engine Optimization and Answer Engine Optimization are real things, even if the terminology is still settling. I’m building frameworks for thinking about them clearly, without the noise that tends to accumulate around anything labelled “AI strategy.”

Structured data, entity recognition, and semantic legibility

What makes a brand or a person legible to a language model? It turns out the answer overlaps significantly with what makes you legible to a traditional search engine — but with some meaningful differences worth mapping.

The Name

2ndvector refers to the second direction: the shift from keyword-based, link-ranked search (the first paradigm) to semantic, AI-mediated retrieval (the second). The lab operates at that inflection point, trying to understand it before anyone has fully figured out what it means.

Current Status

Active research. Writing in progress. Some of it ends up in the News section; the rest is being developed into more structured frameworks.

Not a consultancy. Not taking projects. Just thinking in public.