Elsevier has added a set of AI-powered writing and evidence-checking agents to LeapSpace, its research workspace built on more than 20 million full-text peer-reviewed articles and over 100 million Scopus-indexed scientific records. The expansion, announced on 25 June 2026, targets tasks researchers have identified as high-priority for AI assistance, including manuscript drafting and claim verification across the scientific literature.
The platform already counts thousands of active researchers among its users. Elsevier reports that 97% of them see measurable time savings, with more than half cutting their research time by over 50%. The new capabilities extend that efficiency into writing and evidence synthesis, two areas where speed matters but accuracy is non-negotiable.
What the new agents do
Five new capabilities are rolling out, each designed to handle a specific part of the research workflow while keeping the researcher in control. Writing Coach provides an encrypted drafting environment where researchers can build and refine manuscripts with AI suggestions, approving or rejecting every proposed change. The workspace preserves the researcher's authority over the final text rather than auto-generating content that demands later rewrites.
Claim Radar works as a verification layer. It checks how closely a statement aligns with published literature, surfacing supporting evidence, contradictions, and areas where scientific consensus remains unsettled. Compare Tables pulls evidence from the literature into structured comparison tables based on criteria the researcher defines - useful when weighing competing hypotheses or methodologies side by side.
The update also adds practical workflow connectors. Extended File Upload now accepts Word documents directly, removing the format-conversion step that slows down manuscript work. Reference Export lets researchers push citations straight into their documents, tightening the link between literature review and writing.
How the platform handles evidence
Unlike general-purpose AI tools that generate plausible-sounding text from broad training data, LeapSpace grounds its outputs in peer-reviewed sources and provides traceable citations for every claim. Researchers can inspect the evidence behind any AI-generated suggestion before accepting it. This design choice reflects the high-stakes nature of scientific publishing, where a single unsupported claim can weaken an entire argument.
"LeapSpace now supports even more tasks across the research workflow, and the trusted scientific content it draws on is continually growing as we collaborate with more publishers," said Judy Verses, President, Academic and Government at Elsevier. "Researchers tell us this means stronger scientific arguments, it's quicker to find the evidence they need, and this results in greater confidence - this matters, because the stakes in science are high."
Why this matters for researchers
Time saved on literature searches and manuscript preparation compounds across a career. A researcher who cuts 50% of their research time per project can run more experiments, submit more papers, or spend more time on peer review - the activities that actually advance their field. The writing and verification tools arriving in LeapSpace target precisely the tasks that consume that time: drafting, checking claims against evidence, and organizing citations.
The emphasis on researcher control matters equally. AI that writes without showing its work creates a verification burden - every sentence needs manual fact-checking. An agent that flags its own uncertainties and links directly to supporting papers shifts the burden from detection to judgment, which is where a trained researcher's expertise belongs. For scientists working in fields where consensus is fluid and evidence is contested, that distinction determines whether an AI tool accelerates their work or adds to it.
Your membership also unlocks:
