GatwAI

Scite

Scite shows scholarly citation context and provides a research assistant. Review interpretation limits, current plans, source checks and privacy.

Provider sources checked

What does Scite do?

Scite adds citation context to literature research. Smart Citations classify individual citation passages as supporting, contrasting or mentioning, helping investigate how later work treats a finding. Assistant combines research with composed answers; collections, browser/Zotero integration, MCP and API extend its use. The crucial distinction is between a count and a claim: many citations do not automatically make a study correct. Equally, no recorded contrasting passage does not establish absence of criticism. Classification is a discovery cue leading to the original text. Sound assessment also requires methods, context and coverage checks.
Editorial research · 2026-11-14How we research →

Who is this tool for?

  • Researchers tracing reception of a particular publication.
  • Libraries and editorial teams checking evidence chains and references.

Strengths

  • Citation passages provide more context than counts of citing papers alone.
  • Collections and alerts support ongoing monitoring of a research area.
  • Integrations bring evidence into existing literature and AI workflows.

Limitations and requirements

  • Machine classification and human interpretation can differ.
  • Coverage and licensed full text limit which passages can be analysed.
  • A citation passage is not equivalent to an independent replication study.

Who is behind the product?

Company
Scite LLC, Teil von Research Solutions / part of Research Solutions
Headquarters
Henderson, Nevada, United States: current published contact address.
Company founded
2018 according to its 2019 partnership announcement.
Product launch
Public beta documented by May 2019.

Scite now belongs to Research Solutions. The early citation platform is distinct from later generative assistant capabilities.

Suggested ways to use it

Suggested workflows or verification steps. These are not hands-on tests; for unverified products they are not a recommendation to use the service.

Suggested workflow: challenge a central reference

  1. Open an important publication by DOI or title.
  2. Read supporting, contrasting and mentioning passages with their surroundings.
  3. Record methodological criticism separately from mentions and verify it in the original.

Goal: A nuanced reference note, not a GatwAI classification test.

Suggested workflow: check a draft's evidence

  1. Organize central claims and their assigned references in a collection.
  2. Inspect current citation contexts and applicable editorial notices for each claim.
  3. Flag uncertain attribution and correct the draft after human review.

Goal: A traceable verification path for consequential claims.

Pricing, allowances and data

What should you check before choosing a plan?

Pricing lists Basic at US$12 and Pro at $40 per month with annual billing and a seven-day trial. Basic includes 250 and Pro 2,500 monthly MCP credits; API access and overages differ by plan. Free accounts no longer include MCP credits from October 2026. Enterprise is priced by users and usage.

What should you check about data and permissions?

Terms explicitly exclude training or improving AI with customer data. The policy nevertheless describes processing, providers and retention; account-related personal data may generally remain for ten years after account closure, subject to deletion rights. Shared collections give members file access. Check contractual rules for confidential research separately.

External reviews and available evidence

Bakker and colleagues found low classification agreement in a 2023 sample involving retracted work. A 2025 response by Research Solutions authors challenges differing definitions and the selected sample.

Reported positives

  • The original study and methodological response are linked.

Reported cautions

  • The response has vendor involvement. Both concern older classifications, not a current overall Assistant test.

These reports reflect the linked authors’ experiences. Test date, plan and task affect results; they do not establish a representative overall rating.

Product development and notable changes

  • Collection files and a new Assistant preview

    October release notes add collection files. The new Assistant is initially described as a preview for individual accounts, with additional data types depending on plan.

    Source for this change ↗

Which alternative fits which task?

Semantic Scholar →

Consider for free scholarly discovery and citation connections without the same Smart Citation emphasis.

Elicit →

Consider for tabular extraction of specific study characteristics.

Frequently asked questions

Does a supporting citation prove the claim?

No. It describes a classified passage relationship; study quality and independence require separate checks.

Is MCP permanently free?

No. Updated guidance says Scite MCP requires a paid plan or active trial from October 2026.