New in Relativity aiR: Prompt Kickstarter, Prompt Criteria Validation, and aiR Assist
Relativity continues to enhance its AI-powered review capabilities with three exciting new features in aiR for Review and aiR for Case Strategy. These updates are designed to save time, improve accuracy, and streamline legal workflows by combining advanced AI with structured, defensible processes.
1. Prompt Kickstarter: Rapid AI Prompt Creation
Creating effective AI prompts from scratch can be time-consuming and inconsistent. Prompt Kickstarter transforms your case documents, such as complaints, requests for production, or internal memos, into structured draft prompt criteria in minutes.
By automatically generating case overviews, issue definitions, key terminology, and organizational context, Prompt Kickstarter can cut prompt creation time by up to ten times. This leads to faster onboarding, smoother handoffs between teams, and more consistent review outcomes. Grounded in your actual case materials, AI-crafted drafts provide a strong starting point that your team can refine with legal expertise for better alignment, fewer false positives, and quicker results.
2. Prompt Criteria Validation: Test Before You Apply
Prompt Criteria Validation introduces a deliberate, intermediate test phase in aiR for Review. It allows teams to confirm that their draft review instructions are effective before applying them across an entire document population.
A controlled sample of documents is processed through aiR to generate predictions. Those same documents are then reviewed manually by human reviewers in Review Center. Once coding is complete, aiR predictions are compared to human results using statistical metrics. If the validation metrics meet your acceptance thresholds, the Prompt Criteria can be applied to the full dataset. If not, teams can refine the criteria and repeat the validation process. This feature ensures accuracy and defensibility while reducing the risk of misapplied instructions.
3. aiR Assist: Ask Questions, Get Answers
aiR Assist is a generative-AI feature in RelativityOne that enables users to ask natural language questions about documents in a workspace and receive summarized answers with references. Built on a retrieval-augmented generation (RAG) model, aiR Assist uses indexed document content to provide accurate, context-aware responses.
Users interact through a chat-style panel, posing conversational questions such as:
Who are the key individuals in this case?
How were the Diamond Dogs formed?
aiR Assist returns answers along with citations and links to the documents, giving users an efficient way to explore large datasets, identify key facts, and support case strategy.
Why These Updates Matter
Together, Prompt Kickstarter, Prompt Criteria Validation, and aiR Assist make AI-driven review and case strategy more efficient, accurate, and defensible. Legal teams can spend less time on repetitive tasks and more time on higher-value work while maintaining confidence that AI outputs are grounded in real case materials and validated against human review.
These updates demonstrate how Relativity continues to enhance AI capabilities for modern legal workflows, helping teams move faster, make informed decisions, and maintain defensibility throughout the process.