# Consensus Deep Search (Consensus)

Consensus Deep Search is a standalone vertical AI agent from Consensus. A research agent that plans a literature search across about 200 million papers and writes a cited review of the top 50. Nearfi's Agent Index rates its autonomy 4/5 (multi-step on its own): it breaks a question into a search plan, runs up to 20 searches and returns a cited report with no step approval; it is not 5 because it runs only when asked, and not 3 because the user doesn't approve each step. Status: live; 0 incident(s) logged.

Source: https://nearfi.co/agents/consensus-deep-search

- Category: Research · vertical · Standalone
- What it does: Consensus says Deep Search breaks your question into a multi-step plan, runs up to 20 targeted searches, reviews over 1,000 papers and reports on the top 50, with a consensus meter, key authors, claims and research gaps. An independent university-library review describes it as built on a planning agent, a search agent and a reading agent developed with OpenAI, and typically finishing in under 10 minutes.
- Limits: The index relies mainly on open sources, so full text is mostly limited to open-access papers. The SMU librarian found it hard to trace citations to the exact passage used, said AI-generated filters can be inaccurate, and warned that real sources can still be summarised incorrectly. The launch post carries conflicting dates.