Context
An original preprint introduces a 4B agent-context-aware reranker using the user's original question and available agent reasoning context alongside generated subqueries. In the reported controlled benchmarks it improves dataset-averaged nDCG@10 by 0.9–2.0 percentage points over a base reranker and reduces searches for five tested agents.
What changed
The research tests second-stage document reranking that accounts for a search agent's broader task and reasoning, rather than scoring relevance only against an isolated generated subquery.
Why it matters
Query fan-out alone may not explain which documents are selected for agentic answers. Context-aware reranking adds a distinct source-selection stage that visibility measurement and retrieval experiments should account for.
Visibility OS interpretation
Visibility OS treats this as a demonstrated experimental retrieval mechanism, not evidence that ChatGPT, Gemini or Perplexity currently deploy the method or that optimizing for it guarantees citation gains.
What remains uncertain
Will results replicate after public release of model/code? How does the approach perform with closed-model agents and live search APIs? Does improved reranking translate to citation diversity or source exposure?