Production Set 04 // Vendor Profile, REL
Relativity aiR for Review, capabilities, pricing, honest assessment.
VERIFIED 05 SEP 2026 // INDEPENDENT REVIEW // NOT LEGAL ADVICE
Direct Answer // Relativity aiR Pricing
aiR for Review and aiR for Privilege carry no separate charge. Since 1 November 2025 they are included in RelativityOne under an all-inclusive, unlimited fixed-fee model (announced at Relativity Fest, October 2025). You pay for the RelativityOne platform, not a metered AI add-on.
- -RelativityOne platform: Priced by quote. Relativity does not publish per-GB rates; the drivers are data volume and commitment term (pay as you go vs a one- or three-year commitment).
- -aiR for Review + aiR for Privilege: Included, no separate metered AI add-on. Previously charged as a volume-metered add-on on top of the platform.
- -aiR Assist: Conversational search, standard in RelativityOne from end of June 2026.
- -Billed separately: User licenses (per seat), processing, and hosting (active vs cold tier).
Quote-only // aiR-inclusion model re-confirmed Sep 2026 vs relativity.com/pricing // Confirm rates with Relativity
OCT 2025 // PRICING RESET
Section 01 // Product Stack
What Relativity aiR actually is
Relativity aiR is not a single product, it is a suite of four AI capabilities built into the RelativityOne cloud platform. Understanding the distinction between them matters for procurement and validation planning.
aiR for Review
LLM-scored relevance for document review. The attorney writes a natural-language issue description; the LLM scores documents against it. Works alongside, not replacing, the classical CAL classifier. Launched 2023, matured significantly in 2024 to 2025. It runs on Azure OpenAI (Microsoft's Azure OpenAI Service) under a zero-retention agreement; Relativity does not publish the specific model version, so treat any single named model as unconfirmed.
aiR for Privilege
LLM-assisted privilege detection. The model identifies privilege signals: attorney names, legal advice language, work product indicators, and forwarded-chain privilege propagation. Accuracy is higher than aiR for Review on most document types; Relativity and its partners report strong first-pass privilege detection, though Relativity does not publish a single standardized accuracy benchmark, so treat any specific percentage as vendor- or case-study-reported. The strongest competitive differentiator for aiR versus Everlaw's privilege module.
aiR for Case Strategy
Case-level analytics: document clustering, timeline reconstruction, custodian mapping. This is closer to the ‘agentic’ tier than the other two products. Priced by tailored quote rather than bundled into the standard RelativityOne package as of mid-2026, so treat it as a separate procurement line.
aiR Assist
Conversational, natural-language search over a matter's documents. Built on a retrieval-augmented (RAG) architecture, it takes a plain-language question and returns a citation-supported answer drawn from the underlying corpus. Reached general availability on 16 June 2026 (announced at RelFest London) and became standard in RelativityOne by the end of June 2026. Current scale limits are 300,000 documents per index and 1.5 million per workspace; English-language content only.
Section 02 // Pricing
Pricing after the October 2025 reset
| Component | Pricing (Sep 2026) | Notes |
|---|---|---|
| RelativityOne base subscription | Not published (quote) | By quote: data volume + PAYG vs 1/3-yr commitment |
| aiR for Review | Included (unlimited fixed-fee) | Previously a volume-metered add-on; bundled from 1 Nov 2025 |
| aiR for Privilege | Included (unlimited fixed-fee) | Previously a volume-metered add-on; bundled from 1 Nov 2025 |
| aiR Assist | Included (standard) | Conversational search; standard in RelativityOne from end of Jun 2026 |
| User licenses | Separate (per seat) | Variable by access level |
| Processing fees | Separate | Per-GB ingestion and processing |
| Hosting (inactive data) | Separate (reduced rate) | Cold storage vs active review tier |
Quote-only // Relativity does not publish per-GB rates; aiR-inclusion model and aiR Assist standard-packaging re-confirmed Sep 2026 vs relativity.com/pricing // Confirm with Relativity
Section 03 // Strengths
Strengths
- •Ecosystem depth. The RelativityOne ecosystem (processing, analytics, review, production, redaction in one tenant) eliminates data-movement overhead and simplifies chain-of-custody. The partner network (HaystackID, Consilio, Lighthouse, Epiq) means enterprise matters can be staffed with specialist services without platform switching.
- •aiR for Privilege accuracy. Practitioner reports and vendor case studies consistently rate aiR for Privilege higher than most alternatives on forwarded-chain privilege propagation, the hardest privilege detection problem in typical corporate litigation.
- •Reviewer adoption. The aiR interface is embedded in the existing review workspace; reviewers do not need to change their workflow. AI relevance scores appear alongside the document, not in a separate interface.
- •Post-2025 value. Including aiR at no additional charge substantially improves the economics for high-volume matters, where AI review was previously a separate volume-metered add-on on top of the platform fee.
Section 04 // Weaknesses
Weaknesses
- !RelativityOne only. aiR is not available on Relativity Server. Firms with data-residency requirements, GDPR Article 44+ transfer restrictions on EU data, or on-premises requirements cannot access aiR features.
- !Single model-provider dependency. aiR runs on Azure OpenAI (Microsoft's Azure OpenAI Service), not OpenAI's public API. Documents are processed inside your Azure tenancy and are not retained or used for training, which suits many regulated buyers, but firms with model-provider restrictions (common in financial services, government, and some healthcare organizations) should still confirm their acceptable-use policy permits Azure OpenAI processing of matter documents.
- !Reasoning trace export is partial. Per-document reasoning traces (the LLM's explanation of a relevance determination) are available through a structured JSON API to some Relativity partners, but not to all tenants as a standard export. For matters where the opposing party or the court may require explanation of the AI methodology, this is a gap.
- !Cost at small scale. RelativityOne is priced for enterprise volume and sold by quote, so a single small matter still carries platform, per-seat license, and processing lines. For small-firm, single-matter work a flat-fee review tool is often meaningfully cheaper even with less AI capability.
Section 05 // Fit
Who it is for
Relativity is the right platform for AmLaw 200 and large boutique litigation firms with multi-matter portfolios and existing Relativity workflows, large in-house legal departments that run multiple investigations and regulatory matters in parallel, and enterprise lit-support vendors (HaystackID, Consilio, Lighthouse, Epiq) whose clients expect the Relativity environment. If your firm has ten or more attorneys coding documents on Relativity today, the cost of switching platforms is high and the benefit of the aiR inclusion is real.
Section 06 // Alternatives
When to choose alternatives
- •Small or mid-size firm without existing Relativity workflow: consider Everlaw or Logikcull.
- •Firm with strong deposition prep requirements: Everlaw's Deposition Analyzer is ahead of aiR for Case Strategy at general availability.
- •Firm wanting fully auditable reasoning traces per document: Lighthouse or Nuix Neo are ahead on trace export.
- •Firm with Microsoft or Azure OpenAI vendor restrictions: confirm each alternative's model provider directly. Everlaw leverages leading third-party LLMs and can shift providers rather than committing to one; verify Nuix Neo's current model with the vendor.
- •Single matter under 500 GB: the per-GB economics do not justify Relativity vs Logikcull flat-fee.
Section 07 // FAQ