DistillerSR

Smart Screening reads titles and abstracts and makes inclusion decisions, either by suggesting a call for a reviewer to confirm or by running fully automated on a defined slice of the queue. It plugs into existing forms rather than demanding a new setup, and an independent AI can check its decisions. Evidence Partners points to a Lilly case where AI-assisted title and abstract screening cut review time by 67%, with an accuracy figure of 90% for systematic and targeted reviews. That's the vendor's own result, and the human-in-the-loop mode exists for exactly that reason.

Smart Evidence Extraction, or SEE, pulls data from text and tables at scale. It supports two modes: a human-in-the-loop path where a reviewer confirms each field for regulatory work, and a fully automated path for faster throughput. Anyone who has spent a week copying outcomes from PDF tables into a spreadsheet will recognize the time this addresses.

Every decision in the platform is logged with the reviewer and the time. Version control applies to projects and forms, and the platform is designed around SOC 2 Type II, FDA 21 CFR Part 11, and EU Annex 11. For a CER program, that means the provenance of each data point can be shown on demand. For everyone else, it means a review can be defended after the fact without hunting through old emails.

Projects are built from configurable stages, forms, and filters. You set the protocol once, then standardize it across products or review types by replicating the template. Built-in form libraries cover standard data extraction and risk of bias assessment, including the Cochrane Risk of Bias tool. Filters route references to reviewers based on the answer to a screening question, so nobody sifts through a queue meant for someone else.

CuratorCR is a separate module that turns past extraction work into a searchable repository. When a reference has already been reviewed elsewhere in the organization, the platform says so, shows the data, and identifies who collected it and at which stage. Reviewers validate and reuse it, which removes the most common form of duplicated work in multi-team evidence programs.

References can be imported automatically from corporate library systems and databases, and full texts are available through RightFind or by pulling from PMC. The API connects DistillerSR to BI dashboards, safety systems such as Oracle Argus, and external data sources, so extracted evidence flows into analysis without manual exports. Standard reporting covers PRISMA flow diagrams and activity reports, and reports can be scheduled and emailed to stakeholders.