Enterprise Knowledge Search
Consensus needed to move beyond search so researchers could plan, read, interpret, and connect millions of papers without spending most of their time manually assembling evidence.
Reported impact
Consensus now completes research tasks in minutes instead of weeks, serves more than 8 million researchers worldwide, and increased revenue 8x while preserving a strict standard for cited, low-hallucination scientific answers.
Consensus re-architected its science search product into Scholar Agent, a multi-agent research assistant built on GPT-5 and the Responses API. Instead of generating answers directly from search results, the system first performs context engineering: a Planning Agent decides what evidence is needed, a Search Agent gathers papers from Consensus' index, private libraries, and citation graphs, a Reading Agent interprets papers individually or in batches, and an Analysis Agent synthesizes the findings into the final structured output. Each agent has a narrow scope, which improves discipline and reduces hallucinations, and every answer is backed by a research context pack containing papers, metadata, and key findings. Consensus also tuned the system to refuse unsupported answers when evidence does not meet the quality threshold, then used evaluation pipelines for citation traceability and stylistic consistency so the assistant can grow with new models and new agent types over time.
A researcher question enters Scholar Agent, where the Planning Agent decides the workflow. The Search Agent gathers papers and citation links, the Reading Agent interprets the material, and the Analysis Agent composes a grounded response with a research context pack so every claim traces back to source studies.
"Consensus has expanded rapidly, growing to more than 8 million researchers worldwide and increasing revenue by 8x."
"Every answer comes with a "research context pack""
"If the system can't ground an answer in real evidence, it won't make one up."
"Scholar Agent is a multi-agent system built on GPT-5 and the Responses API."
Review how this reference was scored and trace each normalized claim back to its documented source.
Scoring methodology
Customer Reference Value measures usefulness as a customer reference. Evidence Confidence measures the quality and traceability of available evidence, not independent verification.
Traceability
Field coverage shows which facts each source supports. Dates are shown only when documented by the source.
The official about page says, "What began as a remote-first company, Consensus now calls San Francisco home."
The Organization JSON-LD on the profile lists "numberOfEmployees":{"@type":"QuantitativeValue","value":150}, which maps to SMB under the 0-249 rule.
Related
OpenAI Sample Aug 20, 2026 Stampli Stampli needed to launch complex new products and keep product knowledge current without waiting on scarce design, contractor, and operations bandwidth. 243 modeled active role-hours compressed to about 77 for the Deep Finance launch ChatGPT Work · Codex · Deep Finance OpenAI Sample Jul 30, 2025 Blue J Blue J needed to help tax professionals answer complex questions in a regulated domain where research traditionally requires hours, days, or weeks of reading across statutes, rulings, case law, and commentary, and even… 2.7 hours saved per user weekly GPT-4.1 · Retrieval-Augmented Generation (RAG) system OpenAI Sample Mar 7, 2025 Hebbia Hebbia needed a way to help finance and legal professionals search, interpret, and synthesize huge volumes of private offline documents where standard RAG systems miss implicit answers and manual review takes days or… $2,000 per hour in legal fees GPT-4o · Matrix · OpenAI o1 