Nested Knowledge is an AI-assisted systematic review and evidence synthesis platform built around a workflow called AutoLit, which takes a review team through structured searching, screening, data extraction, critical appraisal, and synthesis inside a single connected tool rather than across separate spreadsheets, reference managers, and word-processor documents. It is aimed at teams producing formal evidence syntheses — systematic reviews, meta-analyses, and scoping reviews — for academic publication, health technology assessment (HTA), and pharmaceutical or medical-device evidence submissions.
This guide explains what Nested Knowledge does, how its AutoLit workflow and AI-assisted features work, what a ‘living review’ is on the platform and how it differs from a one-time systematic review, and what to check before relying on it for a review that will be published or submitted to a regulator or funder.
What Is Nested Knowledge?
Nested Knowledge is a Minneapolis-based company whose core product, AutoLit, is a cloud-based platform for running the full lifecycle of a systematic review or other evidence synthesis project. Rather than exporting search results into a spreadsheet for screening and then into a separate tool for extraction, a review team works inside one shared ‘nest’ (the platform’s term for a review project) from the initial search through to the final synthesis outputs.
The platform positions itself specifically around AI features that are traceable and subject to human review, rather than fully automated black-box outputs — the company states its AI functionality is designed to align with Cochrane’s RAISE (Responsible AI in Evidence Synthesis) recommendations, a set of principles for using AI responsibly in systematic review work that emphasizes transparency, human oversight, and validation over unchecked automation. That framing matters for a reviewer or funder assessing whether AI-assisted output can be trusted in a submission: the relevant question is not whether AI was used, but whether its use was documented and checked by a human reviewer at each consequential step.
The AutoLit Workflow
AutoLit organizes a review into sequential stages, each with its own screen inside the same project:
- Search — structured searching through direct database connections (including automatic PubMed searching with an updatable, saved query), manual record imports from other databases, or targeted search exploration, with automatic deduplication across sources.
- Screening — title/abstract and full-text inclusion/exclusion decisions, run as single or dual (two-reviewer) screening, with an AI-assisted ‘Smart Screener’ that can suggest or triage decisions for a human reviewer to confirm.
- Tagging (data extraction) — extraction against a customizable tag hierarchy, either built from scratch for the project or drawn from a reusable organizational template, with an AI-assisted ‘Adaptive Smart Tags’ feature that proposes extracted values against those tags for review.
- Meta-analytical extraction (optional) — structured quantitative data capture intended to feed a meta-analysis or network meta-analysis.
- Critical appraisal (optional) — risk-of-bias and study-quality assessment against standard tools, with an AI-assisted appraisal aid.
- Synthesis and outputs — automated generation of draft abstracts, manuscript sections, PRISMA-style flow diagrams, and interactive visual dashboards summarizing the included evidence, which stakeholders can explore without needing access to the underlying nest.
Because every stage lives inside the same project, a change made during extraction (for example, discovering during tagging that a study should have been excluded) is easier to trace back through the workflow than when screening and extraction happen in separate, disconnected tools.
What a ‘Living Review’ Means on Nested Knowledge
A living systematic review is a review that is updated on an ongoing basis as new evidence is published, rather than being finalized once and left static until a future full update years later — a model increasingly used for fast-moving evidence bases (COVID-19 treatment evidence being the best-known example, though the model is not limited to any one field). Nested Knowledge supports this pattern in two connected ways.
First, saved database searches (including its automatic PubMed connector) can be re-run against the same query on a recurring basis, with new results flowing into the same nest for screening rather than requiring a fresh project to be built from scratch. Second, its Bibliomine feature can take a PDF of an existing published systematic review or a landmark study and automatically extract its cited references as importable records — a way to seed a new nest from a prior review’s reference list, or to bring a previously published (and now static) review onto the platform as the starting point for converting it into a living review that keeps updating going forward.
Nested Knowledge has also partnered with the citation-analysis platform scite to support living systematic reviews, pairing scite’s citation-context data (how later papers cite and characterize a given study — supporting, disputing, or merely mentioning it) with Nested Knowledge’s review and synthesis workflow. For a review team, the practical implication of a living review is a maintenance question as much as a technical one: someone has to own re-running the search, triaging new records, and deciding when the synthesis and dashboard need to be refreshed — the platform provides the mechanism, not a substitute for that ongoing editorial decision.
Validation and Evidence of Accuracy
Nested Knowledge cites an independent third-party evaluation by Certara (a drug-development and regulatory-science consultancy) that found its AI-assisted screening and extraction steps produced an average workload reduction in the neighborhood of 85% on the review steps tested, while the platform reports this was assessed against accuracy benchmarks rather than speed alone. As with any vendor-cited validation study, a team deciding whether to rely on this for a specific review should look at the original study’s methodology and whether the review type and evidence base it was tested on resembles their own, rather than treating an aggregate workload-reduction figure as guaranteed to transfer to a different disease area or review design.
Who Uses Nested Knowledge
The platform’s stated customer base spans several distinct audiences with different stakes in the output: academic systematic review teams and students producing reviews for publication or thesis work; health technology assessment (HTA) bodies and pharmaceutical or life-sciences organizations building evidence packages for regulatory or reimbursement submissions, where the underlying evidence base often needs to stay current as new trial data emerges; clinical research organizations; and journals or scholarly societies. The HTA and life-sciences use case is a meaningful part of why the living-review and traceability features are emphasized as heavily as they are — a regulatory submission built on stale evidence, or on an AI-assisted extraction step that cannot be audited back to the source document, is a substantially higher-stakes failure than a delayed academic literature review.
What to Verify Before Relying on Nested Knowledge for a Real Review
- Confirm the search strategy is documented and reproducible. Whatever the platform automates, the underlying database search strings, dates run, and inclusion/exclusion criteria still need to be reported in enough detail for PRISMA reporting compliance, exactly as they would for a review run without any AI assistance.
- Treat AI-suggested screening and extraction as a draft, not a final decision. The platform’s own framing — alignment with Cochrane’s RAISE guidance — assumes a human reviewer checks AI-assisted output at each stage; a review that publishes AI-suggested inclusion/exclusion or extracted values without a documented human check undermines the traceability the platform is built to support.
- Check who owns the update cadence for a living review. A living review that is set up but never re-run is functionally a static review with extra infrastructure — confirm a named person or team is responsible for re-running searches and refreshing the synthesis on a defined schedule.
- Verify current pricing and licensing directly with Nested Knowledge before committing a review budget — institutional and project-based pricing for evidence-synthesis platforms in this category changes over time and is not something this guide states as fixed.
Frequently Asked Questions
Is Nested Knowledge free to use?
Nested Knowledge offers a trial and guided demos; ongoing use is a paid, licensed product. Check the vendor’s current pricing directly, since evidence-synthesis software pricing in this category is typically tiered by organization type and project volume and is subject to change.
How is Nested Knowledge different from Covidence, Rayyan, or DistillerSR?
All four are systematic-review screening-and-management platforms with substantial feature overlap, but they differ in emphasis: Nested Knowledge is built especially around connecting the full pipeline (including meta-analytical extraction and living-review updates) inside one workflow, and around HTA/life-sciences use cases specifically. See CASRAI’s systematic literature review tools comparison guide and AI tools for systematic literature reviews guide for a broader look at how the major platforms compare on screening, extraction, and risk-of-bias features.
What is Cochrane’s RAISE guidance, and why does it matter here?
RAISE (Responsible AI in Evidence Synthesis) is guidance from Cochrane on using AI tools responsibly in systematic review work — centered on transparency about where and how AI was used, keeping a human reviewer in the loop at consequential decision points, and validating AI-assisted steps rather than treating their output as ground truth. A platform stating alignment with RAISE is a claim about design intent and process, not a guarantee that any specific review produced on it meets Cochrane’s standards — that still depends on how the review team actually uses the tool.
Can Nested Knowledge produce a full manuscript?
It can generate draft synthesis outputs — abstracts, manuscript sections, PRISMA-style flow diagrams, and dashboards — from the screened and extracted data in a project, but these are drafting aids. A systematic review manuscript intended for journal submission still needs full author review, and typically substantial editing, before submission; see CASRAI’s guide on how to write a research paper for general manuscript-drafting conventions that still apply on top of any AI-generated draft.
For the broader landscape of AI-assisted research tools covered on CASRAI, including other systematic-review and literature-discovery platforms, see the Scholarly Writing for Researchers hub.
