Relativity aiR

Relativity aiR is the generative-AI layer built into RelativityOne for litigation, investigations, and e-discovery. The suite includes aiR for Review, aiR for Privilege, aiR for Case Strategy, and conversational aiR Assist, while claiR provides natural-language interaction inside the platform. The vendor also describes custom extraction and classification tailored to a legal matter. Unlike a general legal chatbot, aiR operates where review teams already host documents, permissions, coding decisions, and productions. That workflow context is its main differentiator. Related legal AI options include CoCounsel, Harvey AI, and GC AI.

Relativity does not publish a fixed dollar amount. Its pricing page says RelativityOne uses flexible licensing with pay-as-you-go terms or commitments of one or three years. It also states that aiR for Review, aiR for Privilege, and aiR for Case Strategy are included at no additional charge with RelativityOne. “Included” does not mean the overall workflow is free: buyers still need a RelativityOne agreement and should confirm data hosting, processing, storage, user, partner-service, and support charges. Request written scenarios for a typical matter, a large production, and a long-running investigation.

The strongest use case is high-volume review where defensibility and supervision matter. Teams can prioritize responsive documents, surface potentially privileged material, ask questions across the record, and build case strategy without exporting evidence to a consumer assistant. Start with a closed matter containing known outcomes. Measure precision, recall, overturn rate, reviewer minutes per document, privilege misses, and time required to explain a result. Sample likely-positive and likely-negative documents; reviewing only the AI’s top results hides false negatives.

Constraints are substantial. aiR is tied to RelativityOne, so it is not a lightweight standalone purchase. Pricing requires sales engagement and may depend on the broader hosting arrangement. Generated answers and classifications are not legal conclusions, and privilege errors can be costly. Teams need documented validation, quality-control sampling, matter-specific instructions, and attorney review. They should verify whether each feature is available in the relevant region and subscription configuration.

A practical rollout begins with one objective, such as identifying a defined issue or prioritizing first-pass review. Freeze the document population, create a human-coded validation set, and compare aiR with the existing workflow. Record instructions and thresholds. Define who can launch analyses, alter prompts, approve coding, and export documents. The vertical AI agents guide adds context for domain-specific systems. aiR is most compelling for organizations already standardized on RelativityOne; others should include migration, hosting, training, and service-provider costs.

Before rollout, document matter-level access, reviewer sampling, privilege escalation, export controls, and a rollback path. Recheck those controls after every model or workflow update.