SciSpace launches review agent
SciSpace introduced an autonomous AI agent that runs end‑to‑end systematic literature reviews from a single prompt. The agent handles PICO questions, screening, data extraction and PRISMA diagram generation — a workflow the company says can shorten projects that normally take 12–18 months. (x.com)
Systematic reviews are the slow, rule-bound studies that ask one narrow research question, search the literature, screen papers, extract results, and document every exclusion. SciSpace says its new review agent can run that workflow from one prompt instead of splitting it across separate tools. (scispace.com) On its Systematic Literature Review page, SciSpace says the agent can search databases, deduplicate records, screen studies, extract data, and produce a report with inclusion and exclusion criteria. The same page says it generates a Preferred Reporting Items for Systematic Reviews and Meta-Analyses, or PRISMA, diagram and follows PRISMA, PRISMA-S, and Cochrane guidance. (scispace.com) SciSpace’s broader agent product was introduced on July 23, 2025 as a research assistant that links “150+” academic tools. The company’s homepage now says the platform covers more than 280 million paper records and can run systematic reviews, draft manuscripts, and match journals from one workspace. (scispace.com, (scispace.com)) The underlying problem is administrative as much as scientific. A systematic review starts with a structured question, often framed as Population, Intervention, Comparator, and Outcomes, or PICO, then moves through screening and data collection steps that Cochrane documents in separate handbook chapters. (cochranelibrary.com, (cochrane.org), (cochrane.org)) PRISMA is the reporting checklist many journals and review teams use to show how records moved from search results to included studies. The PRISMA 2020 guidance says the flow diagram tracks how many records were identified, screened, excluded, and retained, with reasons for exclusion. (prisma-statement.org, (bmj.com)) SciSpace describes its agent as an “AI Co-Scientist” that plans a task, pulls from multiple sources, synthesizes the results, and outputs tables, reviews, and reports. Its prompting guide says the system can search SciSpace, PubMed, arXiv, and Google Scholar, and can also support meta-analysis and data extraction. (scispace.com) The company is also expanding the product around that core review workflow. Its public agent gallery, crawled April 12 and April 13, 2026, listed 883 agents, including separate tools for full-text screening logs, data extraction, grey literature search, risk of bias assessment, and meta-analysis. (scispace.com, (scispace.com)) SciSpace’s pitch is not that review standards have changed, but that one agent can package the same steps researchers already document by hand. If the tool works as advertised, the output still has to satisfy the same PICO framing, screening logic, extraction rules, and PRISMA reporting that systematic reviews have required for years. (cochranelibrary.com, (prisma-statement.org), (scispace.com))
