
DistillerSR
AI systematic review platform for regulated evidence workflows

Tool details
DistillerSR is a systematic review and evidence management platform from DistillerSR Inc. of Canada. Its selling point is not smarter AI but an auditable chain that pins every suggestion to a specific person and time.
Core capabilities
- AI Re-Rank: trains continuously on each reviewer's round-by-round include/exclude decisions and re-ranks the literature; at the 95% recall point it cuts screening burden by a median 47.1%
- Self-trained AI Classifiers: classifiers trained on project data act as a second reviewer and for pre-screening; a 2019 study showed the underlying model was Naive Bayes (87.9%), while SVM reached only 47.6%
- GenAI layer: Smart Screening processes 100,000 abstracts in a single run with no pre-training, and Smart Evidence Extraction traces every point back to the source text
- Smart Chat: launching 2026-10; the LLM reads only the customer's own evidence library and never reaches the open web; included in every edition
- Rules layer: deduplication, forms, RoB, PRISMA, audit logs and LitConnect are not models; add-on modules include CuratorCR, Classifier Manager and the API
Where it fits CER/PER/SOTA for devices and IVDs, PMS/PSUR, HEOR/HTA and pharmacovigilance in pharma, guideline development, and large-sample academic reviews are the best fit. If you only want an AI to read PDFs and answer questions, or you write ordinary reviews alone, it is over-provisioned.
Differentiation Covidence leans toward academic workflows, Rayyan is free but lightweight, and Elicit is strongest at extraction. The difference is audit defensibility: review intensity can be dialed from dual reviewer to fully automated, and SOC 2 Type II, Part 11, EU Annex 11, NIST AI RMF and GDPR are all documented.
Editorial assessment Validation stops at the old 2019–2020 generation: DistillerAI alone reaches only 14% sensitivity and human-machine agreement is κ=0.85; SEE and Smart Screening have no published accuracy figures, and the LLM vendor is undisclosed. The Philips 74%/70% figures come from a 4-person pilot's own account, and the Stryker 70% is a customer estimate whose attribution is data reuse. Data residency is AWS us-east-1 only, which still makes it worth adopting for regulated teams.
Pricing
Subscription$19.95/month (Student tier, 3 users / 1 project)3 users / 1 project, self-serve subscription, includes AI Re-Rank and AI review of exclusion decisions
1 user / 5 projects
1 faculty member + 4 students / 5 projects, includes LitConnect
20–45 projects, includes CuratorCR data reuse and multiple administrators
80–150 projects, includes SSO, AI Classifiers and the API
This page of the official site carries only a lead form, so pricing requires submitting the form
Additional notes
The corporate and government tiers are quote-only with no figures published on the official site; there is no free tier and no self-serve trial, and the official FAQ confirms the free student edition has been withdrawn. Product data residency is AWS us-east-1 only.
FAQ
User reviews
User Reviews
No reviews yet
No one has reviewed this tool yet