Skip to main content
usedby
Search tools, companies

How Consensus uses Zilliz

~45 ms

P99 dense retrieval latency in production across

As published on zilliz.com. Captured by usedby on Oct 6, 2026.

What happened

Consensus uses Zilliz Cloud as the semantic (dense vector) search layer of its research agent platform, searching over 400M+ scholarly sources. The agent calls it as a tool, and the full collection is rebuilt by bulk import into an idle copy of the collection that is then switched live.

Summary written by usedby from the source page, in English. The figures are those of Zilliz and Consensus, not ours.

  • ~45 ms“P99 retrieval at ~45 ms across 400M+ vectors”
  • 14%“14% higher precision in search results after adding semantic search”
  • 27%“the larger embeddings delivered a 27% increase in the quality of papers found”
  • 2.5ד4x larger vector dimensions (256 → 1,024) cost only 2.5× the cluster on Zilliz Cloud”

A daily bulk import of the entire collection turns a full rebuild from a special project into a nightly job. This also unlocks much faster iteration when testing new embedding models.

From the page. zilliz.com, captured Oct 6, 2026

Our job is to make the best research findable for anyone who uses Consensus. Zilliz Cloud gives our research agent fast, high-quality semantic retrieval, directly widening the evidence it can reach.

Christian Salem, Co-founder & Chief Product Officer, Consensus. Source, captured Oct 6, 2026

What the story claims, and what we checked

We compared the story with its live page on Oct 6, 2026.

  • The figure: ~45 msCheckedThe figure is on the page; its label is our wording.
  • The passage quoted aboveCheckedCopied word for word from the page, near the name of Consensus.
  • The numbers in our summaryCheckedEach one is printed on the page.
  • Consensus uses ZillizCheckedConfirmed line. Latest check across sources: Oct 6, 2026.
  • The result itselfNot checkedWe quote it; we did not measure it.

Same company, same tool or same industry.

<200msNeural search latency with Exa Instant, reducedExa uses Zilliz. Another customer of Zilliz60-80%“60-80% of the time spent on consuming, digesting, and identifying relevant data points”Filevine uses Zilliz. Another customer of Zilliz< 10msSearch latency with high recallOpenEvidence uses Zilliz. Another customer of Zilliz