Elicit

Elicit is an AI assistant built specifically for academic research. It was incubated at Ought, a non-profit research lab, and later spun out as its own company. It does not search the general web; it searches scholarly literature. You pose a research question, and it returns relevant papers along with a summary of what those papers say about your question.

Its most valuable ability is not searching but extracting. A normal academic search gives you a list of titles, and you still have to open each one for methods and findings. Elicit can pull study design, sample size, intervention, main findings and similar details out of a batch of papers and lay them out in a table, with each cell traceable to a passage in the source. That is exactly the most tedious, manual part of a literature review.

Be clear-eyed about it, though: extraction is not always accurate, particularly for numbers and fine details, and anything going into a review must be checked against the original. Its corpus is also mainly English-language, with little coverage of Chinese journals, so it suits researchers and students working with international literature.

Key features

• Find papers by question: describe a research question in plain language and get relevant papers plus a synthesised answer from the top results.
• Bulk extraction: define the fields you want, such as methods, sample and conclusions, and compile many papers into a comparison table.
• Source tracing: extracted content links back to the relevant passage in the paper so you can verify it.
• PDF upload: bring in papers you already have and include them in extraction and comparison.
• Systematic review workflow: a more structured screening and extraction flow to support the early stages of a systematic review.

How to use

1. Create an Elicit account; the free credits are enough to run the full workflow a few times.
2. Write your research question as one complete sentence in English. The more specific it is, the more relevant the papers.
3. Go through the results, keep the papers that are genuinely relevant and drop the ones that clearly are not.
4. Add the columns you want extracted, generate the table, and export it to a spreadsheet for further work.
5. For any finding going into a paper or report, click back to the original and check it.

Best for

• Starting a literature review: quickly mapping the key studies on a research question.
• Comparing evidence: lining up design, samples and results across several studies.
• Evidence-based decisions: finding out quickly whether research supports a given practice.
• Thesis writing: helping students organise related work instead of searching aimlessly.

Strengths and limitations

Strengths

• Bulk extraction into tables is genuinely useful and saves a great deal of reading and note-taking.
• Results trace back to the source, which suits academic work better than a general chat model.
• Answers draw only on scholarly literature, keeping out marketing pages and other low-quality sources.

Limitations

• Extraction has errors and omissions; numbers and details must be checked by hand.
• The corpus is mostly English, with almost no coverage of Chinese journals.
• Some papers are available only as abstracts; full-text extraction depends on open access or your own uploads.
• Free credits are limited, and large reviews need a paid plan.