Consensus (consensus.app) is designed for one specific task: answering yes/no or “what does the evidence say about X” questions by searching and synthesizing actual peer-reviewed research. It doesn’t write essays or generate content — it finds studies and tells you what they found.
How Consensus Works
Ask a research question (“Does exercise reduce cognitive decline in adults over 60?”) and Consensus searches its database of peer-reviewed papers, returns the 10–20 most relevant studies, and gives an “Consensus Meter” — a breakdown of studies that found yes, no, or mixed results on your question. Each study shows a one-sentence finding extracted from the paper.
The “Study Snapshot” feature shows methodology details — RCT vs observational, sample size, follow-up duration. This matters because a systematic review of 50 RCTs weighs differently from a single observational study of 200 people.
Best Use Cases
Hypothesis testing before designing a study: checking if your proposed intervention has already been tested and what the evidence shows. Writing introduction and discussion sections: finding what the field currently believes about your topic. Grant writing: demonstrating evidence gaps that justify your proposed research. Clinical researchers: checking whether a treatment effect found in one paper replicates across literature.
Limitations
Consensus works best for health, social science, and education research — areas with lots of well-defined RCTs and clear outcome measures. It’s weaker for physics, chemistry, and engineering where research questions are less binary. The database is large but not complete; very recent papers (less than a year old) may not appear.
Comparing to Other Tools
Where Perplexity gives you an answer and Elicit gives you a table, Consensus gives you a verdict with evidence quality context. For questions with a binary character (“does X work?”), Consensus is often the most efficient first tool. For open-ended questions (“what are the mechanisms of X?”), Elicit or Semantic Scholar is better.
