Consensus

Consensus is an AI-powered search engine focused on scientific research papers. Instead of general web pages, it searches across millions of academic studies and helps you understand what the literature says about a given question. You query it in plain language, and it returns peer‑reviewed sources with AI-generated summaries and context.

The tool is aimed at researchers, students, clinicians, and organizations like universities and research groups, but it’s also pitched to “citizen scientists” and anyone outside academia who wants to access real evidence. Example prompts on the site span medicine, psychology, economics, AI, education, and more, showing it can support everything from literature review outlines to comparison tables and evidence-based blog drafts. You access Consensus through the web, and can start by “searching for free”; there are also references to a Pro experience and offerings “for Universities” and “for Organizations.”

Consensus works best as an evidence discovery and synthesis assistant, not as a fully autonomous research agent. According to user testimonials on the site, it “doesn't write for me, but it sets me up to write,” by highlighting the most relevant papers, summarizing them clearly, and identifying study designs. It can help you draft outlines, tables, and structured comparisons, but you’re still responsible for critical appraisal, writing, and interpretation.

The site does not detail which AI models it uses, how often the research index is updated, or specific privacy guarantees, so those technical aspects remain unclear from the public content. It also isn’t described as integrating with external tools, APIs, or workflows, so if you need an agent that autonomously executes multi-step research tasks across systems, this likely won’t replace a more general agent platform. Pricing details for Pro or organizational access are not publicly listed; you can search for free and are directed to contact them for institutional use. It’s a good fit if you want faster, more structured access to peer‑reviewed evidence, but not if you need automated experimentation, code execution, or deeply customized agent behavior.