Q3 Product Roadmap – A Letter from the Product Team

Welcome to Nested Knowledge’s first ever prospective product development post! This has been long overdue. Historically, we’ve pushed release notes as they occur, and careful listeners maybe gleaned upcoming features from our videos, blog, and trainings. As usership, product capabilities, and AI possibilities grow, we want researchers to know where NK is headed.

Mission Statement

Nested Knowledge is building widely accepted SLR workflows, automating them, and providing human oversight on those automations. We will accommodate our expert users no matter their review protocol and where they fall on the AI acceptability & optimism spectrum.

Q3 Conceptual Goals

• Automate SLR: Coordinate & improve upon existing Smart features.

• Complete SLR Workflows: Enable an even larger number of review protocols to be accomplished within AutoLit.

• Data Integrations: Provide & ease access to data sources researchers are accustomed to.

• Move Downstream: How can NK make review data & synthesis more explorable & useful in downstream applications?

Q3 Feature Development

• Automations
  • Smart Nest: We have automations in place at every step of an SLR. All that’s left is to tie them together into one process. The user will provide a research question & instructions, and a rapid review across thousands of records will be generated for them within minutes.
  • ASTs & Smart Screener improvements: LLM version bumps, caching, and pre/post processing steps will improve the quality and speed of extraction automations.
  • Supplemental Materials: ASTs can be optionally enabled to extract from Supplemental Materials in addition to Full Texts of records.

SLR Workflows:

• Dual Criteria-Based Screening: 2 reviewers screen records independently before adjudication of both criteria and screening decisions.

• Reviewer-level Smart Screener: Replace one of the human reviewers with Smart Screener.

• Auto-Adjudication: For records where reviewers agree, automatically provide final criteria and screening decisions. The definition of agreement, particularly in the case of NR decisions, is configurable by the adjudicator.

• Within-nest Smart Screener Evaluation: A diagnostic & debugging tool comparing Smart Screener with human criteria decisions.

• Adjudicated Tagging: Either 1 or 2 reviewers extract data independently before another reviewer adjudicates

• Reviewer-level ASTs: Replace one of the human reviewers in Adjudicated Tagging with ASTs. Human oversight is provided via the adjudicator.

• Advanced Search: Build search strategies with line-by-line decomposition, errors, comments, and result counts.

Data:

• Mum’s the word: We are in talks to integrate several external databases or platforms, but are not able to disclose them quite yet! 
• Configured Usage Rights: Maintain an organization-wide database of abstract and full text usage rights, applied automatically to all nests the organization owns. Abstract usage rights will be managed on a search engine basis, while full text usage rights will be managed at the journal (ISSN) level.

• Synthesis Chat: In a chat interface, ask questions, analyze, and generate reports on evidence contained within your nest (records, screening decisions, extractions, appraisals).

• Smart Abstract: Summarize the methods and key results of your review for Synthesis.

Beyond

Looking ahead to Q4 & Q1 27, we are tentatively planning:

• Quantitative Extraction v2: More diverse types of extractions and meta-analyses than supported in MAE v1. Improved Smart MAE. Quantitative Insights. We expect to build these capabilities on top of Tagging: Tables & Index Tables.
• Related Report Management: Automatically detect & structure multiple reports of the same study for unified extraction; avoid duplicative evidence reporting & synthesis. 

• Improved documentation: layout, formatting, search, interactivity (Ask AI v2).

Closing

“In 7 years of developing Nested Knowledge, I have never felt SLR methodology shifting so rapidly. It’s challenging & invigorating for our team, and possibly you too! We aim to keep researchers on the leading edge of SLR methods developments while continuing to provide all workflows they are accustomed to.”

-Karl Holub, CTO Nested Knowledge