Our Take
Relativity is embedding AI query and custom review logic directly into its core platform rather than bolting it on as a separate layer, which matters because defensibility and auditability in legal work require the same system to be the source of truth for both data and analysis.
Why it matters
Legal teams run discovery and review on platforms, not in separate AI tools. If RelativityOne now handles both the vault and the analysis in one environment, practitioners no longer need to export data, lose audit trails, or manage multiple systems for evidence and reasoning.
Do this week
Legal ops: test aiR Assist on your next matter intake before July 1 so your team can measure whether plain-language queries cut the time your reviewers spend building search strategies.
Relativity ships aiR Assist as a standard feature
Relativity announced general availability of aiR Assist and custom analyses in aiR for Review, both rolling into RelativityOne as standard offerings by end of June (company-reported). The feature set comes shortly after Relativity acquired contract AI firm Gavel.
aiR Assist lets legal teams ask plain-language questions of document data and receive cited answers instantly. Custom analyses extend aiR for Review by allowing teams to define, deploy, and scale document review workflows in plain language without coding. The platform handles up to 300,000 documents per index and 1.5 million documents per workspace (company-reported).
Phil Saunders, Relativity's CEO, framed the move as a difference in architecture: almost any AI system can read a document, he said, but few can serve as the single auditable source of truth for every piece of data in a matter. The distinction matters in court and before regulators, where outputs must be defensible by design.
Auditability and governance are built in, not added later
Legal AI deployments fail when teams use separate AI tools for analysis and then copy results back into their case management system. The handoff introduces risk: gaps in chain of custody, lost context, reasoning that cannot be traced back to source data. Regulators and opposing counsel both exploit these gaps.
By embedding aiR Assist and custom analyses inside RelativityOne rather than treating them as external services, Relativity removes the export-and-import step. The same system that stores and indexes documents also reasons over them. A single audit log covers both the data and the analysis. That changes the compliance calculus, especially for teams running large document sets where manual spot-checking is impossible.
The plain-language query model also lowers the barrier for non-technical reviewers. Discovery teams no longer need to wait for a technical analyst to write Boolean searches or configure workflows; they can ask questions directly and iterate.
How legal teams should approach this
Start with a pilot on a current matter where your team is already in RelativityOne. Do not assume all use cases benefit equally. Document review on well-defined issues (contract classification, privilege identification) will show faster value than open-ended fact discovery, where human judgment still dominates.
Test the auditability claim before you commit. Run a small batch of documents through custom analyses, export the outputs with their reasoning traces, and ask your opposing counsel's tech team to reproduce a handful of results. If the audit trail holds, the system has real defensibility; if it does not, you have learned the scope before scaling.
Relativity's architecture removes friction at the system level, but it does not eliminate the need for quality control. Plan for human review of aiR Assist outputs, especially on novel matters. The tool finds facts faster; it does not substitute for strategy.