ASReview
asreview.nl
- Free plan
- Open source
- Windows
- macOS
- Linux
- Web
- API
- Self-hosted
What ASReview is
ASReview is a free, open-source tool that helps prioritize records for systematic review screening with AI-aided active learning, while a human screener retains the decision. The site says the approach can reduce screening workload by up to 95%. Users can simulate AI-aided reviewing to compare models, and teams can screen in parallel with Crowdscreen, where administrators manage access through private login. The model series described include ELAS-Ultra, ELAS-Heavy, and language-agnostic ELAS-L; the latter two require the Dory package. ASReview supports reference managers including EndNote and PubMed, plus CSV, RIS, and XLSX import and export. Documentation covers a Python API, while institutional workflows also describe a REST API and CLI for automation. In a local setup, data stay on the user's computer; server deployments keep data within the organization's environment and policies. The project provides documentation, tutorials, videos, and community discussion, as well as workshops and custom consultancy for organizations.
Who it suits
It suits screeners, information specialists, students, and institutions managing systematic reviews. Developers and data scientists may also find its APIs and automated workflow options relevant.
Good to know, and what to watch
- Human screeners retain decision-making
- Simulation supports model comparison
- Crowdscreen enables parallel team screening
- Imports and exports CSV, RIS, and XLSX
- ELAS-Heavy and ELAS-L require the Dory package
- AI paper summaries are not supported
- Full-text management is not supported
Plans
Every plan and priceASReview LAB
Free and open-source ASReview LAB
Free plan
Getting started
- Open the ASReview website
- Choose a local, server, or self-hosted deployment
- Import records from CSV, RIS, or XLSX
- Select an available AI-aided screening model
- Run screening with human decisions or use Crowdscreen for parallel team work
- Export records in CSV, RIS, or XLSX format
Limits to check
ASReview LAB is free and open source with no licensing fees or subscriptions, but running your own hardware or server may incur costs. ELAS-Heavy and language-agnostic ELAS-L require the Dory package.
The verdict
ASReview combines AI-aided prioritization with human screening decisions and options for individual or team workflows. Its local and server deployment choices offer different ways to keep review data within a user's or organization's environment.
Questions people ask
Is ASReview free?
Yes. ASReview LAB is listed as free, with no licensing fees or subscriptions.
Which platforms are supported?
Platforms listed include API, Linux, macOS, self-hosted, web, and Windows.
Can teams screen together?
Yes. Crowdscreen supports parallel screening, with administrators managing access through a private login system.
What files can it import and export?
The institutional workflow supports CSV, RIS, and XLSX import and export. ASReview also describes compatibility with EndNote, PubMed, and other reference managers.
Where is review data kept?
In the default local setup, data stays on the user's computer. Server deployments keep it within the organization's environment and apply that organization's policies.
Does it offer an API?
Documentation includes a Python API, and institutional workflows mention a REST API and command-line interface for automated work.
What the maker publishes
- AI models
- ASReview describes ELAS-Ultra, ELAS-Heavy, and language-agnostic ELAS-L model series; the latter two require the Dory package. source
- AI models
- Version 2 presets include ELAS-ultra, ELAS-heavy and ELAS-lang, with configurable classifiers, feature extractors and query strategies. source
- API
- ASReview documentation includes a Python API, and the institutional page also mentions a REST API and CLI for automated workflows. source
- API maturity
- The API documentation warns that the API is still under development and may change without warning. source
- Citation-manager integrations
- Documented citation-manager compatibility includes EndNote, Excel, Mendeley, RefWorks and Zotero through supported export formats. source
- Company description
- LinkedIn describes ASReview as open-source software developed at Utrecht University for active-learning screening of scientific papers, policy reports and social-media data. source
- Company headquarters
- LinkedIn lists Utrecht, Netherlands as the headquarters and primary location. source
- Company size
- LinkedIn lists the organization size as 11–50 employees. source
- Crowdscreen
- Crowdscreen supports parallel screening by a team, with administrators managing access through a private login system. source
- Crowdscreen
- Crowdscreen lets AI agents propose records to a crowd of experts. source
- Custom models
- Model extensions can add classifiers, query strategies, balancing strategies and feature-extraction methods. source
- Dataset quality requirement
- Official documentation says models work best when duplicates are removed and titles and abstracts are complete for most records. source
- Documentation depth
- The documentation covers user, server and technical guides, including authentication, command-line use, APIs and extension development. source
- Documentation support
- The maker directs users to more than 500 pages of documentation and a GitHub community for help. source
- Double-screening limitation
- The independent review describes ASReview as a single-screener prioritization engine without multi-reviewer blind screening. source
- Extension system
- Python extensions can add model, subcommand and dataset functionality through setuptools entry points. source
- Formats
- The institutional workflow page describes CSV, RIS, and XLSX import and export. source
- Full-text limitation
- ASReview focuses on title-and-abstract screening and does not provide downstream full-text review or data-extraction workflows. source
- Headquarters
- Utrecht, Netherlands source
- Human decisions
- Human reviewers label records while the model continuously updates the ranking. source
- Independent praise: auditability
- RSRCHDB describes the open code and peer-reviewed benchmarks as useful when screening decisions must survive audit. source
- Independent praise: privacy
- The Augmented Scholar highlights that local operation requires no account or upload and keeps unpublished review data on the user’s machine. source
- Independent praise: transparency
- The Augmented Scholar praises the open algorithms, versioned software, peer-reviewed method and ability to benchmark simulations locally. source
- Integrations
- The site describes compatibility with EndNote, PubMed, and other reference managers, and says RIS files can be imported and exported. source
Checked against the maker's pages on 15 Oct 2026.