BLURSOR — AI Research for Practitioners

In March 2026, researchers published over 2,800 papers on how AI decides which sources to trust. That’s 70,000+ pages in a single month.

Are you reading this?

Neither are most practitioners.

Not because they don’t care. Because the volume is impossible.

We are closing that gap.

How Blursor Works

BLURSOR article — citation failure taxonomy

We monitor research hubs for papers on AI visibility, LLM ranking factors, and how generative engines decide what sources to cite.

We filter for what matters in practice. We extract the findings that have real implications and distill them into readable, actionable articles.

Then we publish — the insights, the supporting data, and what it means for your work.

The original paper is always linked.

↓ This is what it looks like

Citation · Geography

Diagnosing Citation Failures in Generative Engine Optimization: What AgentGEO Actually Shows

A new paper from Virginia Tech maps four failure modes that prevent pages from being cited in AI-generated responses. 43% of relevant pages receive zero citations under baseline conditions.

May 22, 2026 · 7 min read · arXiv:2603.09296

All articles →