aiR Review (formerly known as aiR for Review) is a generative AI application integrated into Relativity aiR that uses large language models to review documents and support coding decisions with explainable rationale.

Key features and functions include:

Prompt-based document analysis
Uses natural-language "Prompt Criteria" defined by the user — based on a review protocol — as the basis for analyzing each document, allowing teams to iterate on criteria and refine results over successive runs.

Three analysis types
Supports Relevance Review, Issues Review (up to five issues per run), and Key Document Identification, selected per aiR Review job.

Explainable predictions with citations
Generates a prediction for each document along with a rationale, considerations, and document excerpts supporting the decision, viewable in an analysis pane within the document Viewer.

Quality control for traditional review
Compares aiR Review's coding predictions against decisions made by human reviewers, intended to accelerate QC and surface discrepancies for further evaluation.

Fast Track and Batch processing modes
Offers two processing modes per job to accommodate different review speed and volume needs.

Internal investigation support
Helps surface documents and insights relevant to understanding the underlying facts of a matter, beyond standard relevance coding.