Elicit is an AI research assistant built for scientific literature review. It searches tens of millions of papers, finds the most relevant ones and, most importantly, lays out each paper's information (method, sample size, main result, limitations) side by side in a table. It can produce an automated research report and, in systematic reviews, speed up the screening and data extraction steps. What sets it apart from competitors is its focus on this "extraction table", which a researcher would usually spend days building by hand in a spreadsheet.

What is Elicit good for?#

  • Thesis literature review: a management PhD student finds papers on "remote work and productivity" and sees each one's method and result in a table.
  • Systematic review: a medical researcher screens papers against inclusion and exclusion criteria and extracts data.
  • A quick research report: for a specific question, you get a structured report with references to the papers.
  • Chatting with a paper: open a long paper and ask about its statistical method or limitations.
  • Finding research gaps: by viewing findings side by side, you find areas that are less studied.

Strengths and weaknesses#

Strengths:

  • The data-extraction table genuinely saves a researcher's time.
  • Each data point links back to the paper's original text for careful checking.
  • Suited to systematic review methodology.

Weaknesses:

  • The free plan is limited for serious work and the professional subscriptions are expensive.
  • Extracted data is sometimes wrong or incomplete and must be checked one by one.
  • Coverage of the humanities and non-English sources is weak.

Free version and pricing#

The free plan offers search, summaries and chat with a paper, but the number of automated reports and extraction-table columns is limited. Paid subscriptions for individual researchers and teams raise these caps and unlock systematic review features. Check the official site for current prices.

Language support#

Elicit is built for English-language papers; it understands questions in other languages to some degree, but it doesn't cover papers in other languages, so it is better to write your question in English. You can translate the table output afterwards with a chatbot.

Getting started in three steps#

  1. Write your research question precisely and in English.
  2. From the results, select the relevant papers and add the columns you need (such as sample size or result).
  3. Check the extracted data against the paper's original text, then use it in your work.

Elicit alternatives#

  • Consensus: for a quick answer to a scientific question and seeing how much research agrees.
  • SciSpace: helps with reading and explaining hard parts of a paper and with writing scientific text.
  • NotebookLM: upload selected papers and ask questions of them directly.

For working with long texts, also see Working with long documents using AI: summaries, questions and comparison.