Elicit

Elicit is a specialist research product, not a web answer engine with an academic skin. Its useful unit of work is a literature-review step: finding papers, screening them, extracting structured fields and preserving enough provenance for a researcher to inspect the evidence.

That specialization is also its boundary. A large paper index and an AI-generated summary do not guarantee exhaustive recall, valid study interpretation or a defensible systematic review. The researcher remains responsible for protocol, databases, exclusions and final synthesis.

> Distinctive strength: A paper-centered workflow that connects discovery, screening and extraction. > > Where it stops being an advantage: It cannot turn an incomplete search strategy or weak eligibility protocol into reliable evidence.

Decide in 60 seconds

Price the review, not the query

The reviewed pricing page lists Plus at $11 per month on annual billing, Pro at $39 annually billed or $49 monthly, and Scale at $89 annually billed or $169 monthly. Each tier changes workflow allowances. Comparing them by “AI searches” alone misses the expensive steps: screening volume, extraction, reports and systematic-review projects.

For a team with four active reviews, calculate:

`cost per completed review stage = subscription + exported/manual work + verification labor`

Count papers retrieved, screened, included and extracted separately. A plan that retrieves more irrelevant papers can increase total labor even when its subscription is cheaper.

A recall-and-extraction benchmark

Choose one completed review for which the team already has a verified included-paper set. Freeze the question, eligibility criteria and search date. Without feeding Elicit the answer set, reproduce discovery and screening.

Measure:

• how many known included papers are retrieved in the first 50, 100 and full reviewed set;
• new plausible papers the original review missed;
• false exclusions during title/abstract screening;
• extraction accuracy for five fields with different difficulty;
• whether every extracted value links back to inspectable source text;
• researcher minutes to resolve discrepancies.

This test reveals whether Elicit saves review work in your domain. A generic demo question cannot.

Method and governance boundaries

Document the databases Elicit searches, field coverage, language limits, preprints, date lag, deduplication and export format. For regulated or confidential research, verify account roles, retention, deletion, subprocessors and contractual controls. Keep the search protocol and exclusion reasons outside any single vendor so the review remains reproducible.

Do not rely on Elicit alone if

• the review requires exhaustive multi-database retrieval;
• the field depends heavily on sources outside scholarly papers;
• inclusion decisions cannot receive independent human review;
• extracted values will be published without checking the underlying paper.

Alternatives by evidence task

Elicit verdict

Elicit is a strong shortlist candidate for teams that can define a literature-review protocol and measure recall. It should augment—not silently replace—the accountable research method.

Primary action: Review Elicit's current plans ↗, choose one completed review with a known paper set, and run the recall-and-extraction benchmark before moving an active project.

Official sources checked

• Elicit plans and workflow allowances ↗
• Elicit product documentation ↗
• Elicit privacy policy ↗
• Elicit industry workflow for evidence and policy research ↗

Sources checked 2026-09-02. Confirm current limits and institutional terms before procurement.