Litigation support teams have moved from manual Concordance review to AI-native platforms. Here is what they are running today — and why the stack has changed so much since 2020.
Litigation support teams use eDiscovery platforms for document review — the most widely deployed being Relativity (and RelativityOne), Everlaw, CS Disco, and DecoverAI for AI-native review. These platforms handle the full discovery workflow: data ingestion and processing, first-pass responsiveness review, privilege review, redaction, and production delivery — with AI increasingly automating the steps that once required large contract review teams.
The specific mix of software a given litigation support team runs depends on matter size, budget, and how aggressively the team has adopted AI. A large Am Law 100 firm running multi-million document matters might anchor on Relativity with certified admins and a TAR workflow. A boutique or in-house team doing a 50,000-document investigation increasingly goes straight to an AI-native platform like DecoverAI, skips the admin overhead, and gets from ingestion to production in days instead of weeks. The stack has never been more heterogeneous — and never more consequential to get right.
Modern litigation support teams run four functional layers of software. Understanding where each layer begins and ends is essential for evaluating vendors and managing costs. In 2020, lit support teams routinely managed five to seven different software tools to cover these layers. In 2026, modern all-in-one platforms handle ingestion through production in a single environment, eliminating most of the handoff errors that used to consume QC time.
Review platform. The review platform is the center of gravity — it covers roughly 80% of what a litigation support professional does each day. It stores the document universe, manages user access, tracks coding decisions, enforces privilege, and outputs the production. Relativity, Everlaw, CS Disco, and DecoverAI are the most commonly encountered review platforms in 2026.
Collection tools. Before documents reach a review platform, they must be collected from custodians. For enterprise O365 environments, Microsoft Purview (formerly the M365 Compliance Center) provides native legal hold and export without third-party tooling. Google Vault serves the same function for Google Workspace. For mobile devices — iOS and Android — Cellebrite remains the dominant forensic collection tool used by litigation support vendors and law enforcement alike. Nuix Investigate is widely used for complex enterprise collections that require chain-of-custody documentation and processing in a single workflow.
Processing. Processing converts raw collected data — PST files, loose emails, SharePoint exports, Slack exports, cloud drive packages — into reviewer-ready documents with extracted text, metadata, near-duplicate identification, and email threading. Modern cloud platforms have absorbed this step natively: Everlaw, Disco, and DecoverAI all process documents on ingest, eliminating the need for a separate processing tool in most matters. Teams running legacy Relativity workflows may still use Nuix or Ipro for processing before loading into Relativity, though this is an increasingly unnecessary step for new matters.
Communication and matter management. Litigation support does not live in review platforms alone. Teams coordinate custodian interviews, collection timelines, review batch assignments, and production deadlines across Microsoft Teams or Slack for day-to-day coordination, and dedicated matter management systems like TeamConnect, LegalFiles, iManage, or Clio for docketing and file organization. The integration between the review platform and the project management layer is often an afterthought — but in large matters with multiple review vendors, a weak handoff between systems is where errors occur.
Below is a detailed breakdown of each major platform — its architecture, pricing model, where it fits best, and what trade-offs a litigation support professional should understand before committing to it for a matter.
The dominant platform in the Am Law 200 and among national eDiscovery vendors, Relativity was built on a .NET stack and grew out of the Chicago eDiscovery market. RelativityOne is the current Azure-hosted cloud version; it is where most new workspaces are created today. Relativity's core strengths are its depth of customization, the size of the Relativity Certified Administrator (RCA) credentialing ecosystem, and its Technology-Assisted Review capability called Active Learning. For Am Law 100 firms and the major litigation support vendors — Epiq, Consilio, HaystackID — Relativity is the default platform around which their entire operational workflow is built.
Everlaw is a cloud-native platform built in the 2010s for the post-Relativity world. Its interface is notably cleaner than Relativity’s, and it has earned strong adoption in litigation boutiques and mid-size firms where the review team often includes case attorneys directly rather than just contract reviewers. Everlaw’s predictive coding and AI tagging features are competitive, and its timeline, storyboarding, and deposition preparation features are genuinely differentiated in the market.
Disco was built specifically for corporate legal departments and law firms handling commercial litigation. Its AI assistant, Cecilia, provides AI-powered search, concept clustering, and document scoring out of the box. Disco went public in 2021 and has since refocused its positioning around corporate legal departments managing repeat discovery obligations, where its modern interface and strong analytics capabilities are well-regarded.
DecoverAI is the AI-native review platform built specifically after LLMs became commercially viable — designed for the post-ChatGPT review paradigm rather than retrofitted with AI as a feature layer on top of a legacy architecture. The platform runs multiple commercial LLMs simultaneously against the document set and reconciles their outputs, treating consensus as signal and divergence as a flag for attorney review. Privilege log generation, PII redaction, and production delivery are first-class features, not add-on modules.
Nuix is the processing and investigation workhorse of the eDiscovery world. It is less a review platform and more a data processing and forensic analysis engine. Law enforcement agencies, government regulators, and corporate investigation teams use Nuix to process raw data, carve artifacts from disk images, and build investigation datasets. Nuix Investigate adds visualization and analytics capabilities, including communication pattern mapping that remains distinctive for complex internal investigations.
The platform choice ripples through the entire matter budget. Per-user pricing models penalize teams that need to scale reviewers quickly. Per-GB all-in pricing rewards efficiency and AI automation. Before selecting a review platform, calculate the total cost including admin overhead, processing fees, user seats, and production charges — not just the headline rate. On a 100 GB matter with 10 reviewers over six months, the difference between a per-user platform and a per-GB platform can exceed $100,000.
| Platform | Pricing model | AI review | Privilege log auto-gen | No seat fees | Admin required |
|---|---|---|---|---|---|
| Relativity / RelativityOne | Per-user + per-GB | Active Learning (TAR) | Manual | No | RCA required |
| Everlaw | Per-user + per-GB | AI tagging + TAR | Semi-manual | No | Low overhead |
| CS Disco | Per-matter subscription | Cecilia AI | Semi-manual | Varies | Low overhead |
| DecoverAI | $60/GB/month all-in | Multi-model LLM | Automated | Yes | None required |
| Nuix | Enterprise license | Processing only | N/A | N/A | Specialist required |
Even teams anchored on legacy platforms are bolting AI capabilities onto their workflows. The categories where AI is having the most measurable impact in 2026:
AI first-pass review. The traditional first-pass review model — batch documents to contract reviewers who code responsive/non-responsive at $35–$50 per hour — is being replaced by AI classification that scores the entire document universe in hours. The AI identifies a responsive core, surfaces the highest-value documents first, and reduces the number of documents requiring human review. On a 200,000-document matter, this can reduce billable review hours by 60–75% without sacrificing recall accuracy. Platforms like DecoverAI use multi-model consensus specifically to push recall above 95% before any human review begins.
AI privilege detection. Attorney-client privilege review has historically been the most labor-intensive step in document review. AI privilege detection identifies documents containing legal advice, marks attorney names and law firm domains as privilege indicators, and flags near-certain privilege documents for attorney confirmation rather than full attorney review. The downstream benefit is a substantially reduced privilege review queue and a privilege log that assembles automatically from the coded population.
AI privilege log drafting. Instead of a paralegal manually typing a description for each of 2,000 withheld documents, AI drafts the privilege log entry — summarizing the nature of the communication, the communicants, and the applicable privilege basis — and the attorney approves, edits, or rejects each entry. What used to take 20–40 hours of paralegal time takes 2–3 hours of attorney QC time.
Auto-redaction for PII. Data breach litigation, healthcare matters, and any case involving consumer data requires redacting personally identifiable information before production. AI redaction tools identify Social Security numbers, credit card numbers, protected health information, dates of birth, and financial account numbers at scale. Automated redaction covers all standard PII categories and presents proposed redactions for attorney confirmation before burning them into the production set.
Translation and foreign-language review. Cross-border investigations involving documents in Mandarin, Spanish, German, Korean, or Arabic have historically required bilingual contract reviewers or batch translation services — both slow and expensive. AI translation integrated directly into the review platform lets reviewers see machine-translated versions of foreign-language documents in real time, with the original preserved. This does not eliminate the need for attorney review of key foreign-language documents, but it eliminates the bottleneck of waiting for translation batches before review can proceed.
Communications analytics. Internal investigations — especially those involving potential fraud, harassment, or regulatory violations — benefit from AI tools that surface unusual communication patterns, identify key custodians based on network centrality, and flag sentiment anomalies that warrant closer review. Nuix Investigate and Brainspace are the most established tools in this category; some review platforms now include communications analytics natively.
How lit support teams deploy these capabilities depends on their platform choice. On DecoverAI, all of the above are included in the $60/GB/month rate. On Relativity or Everlaw, most require either additional modules, third-party integrations, or manual process substitutions — each adding cost and complexity to the matter workflow.
The shift in litigation support technology since 2020 is as significant as any five-year period in the industry’s history. The comparison below captures what a typical commercial litigation team was running in 2020 versus what a modern team runs today.
Three forces drove this change. LLM capabilities crossed the threshold for document review accuracy in 2023–2024 — the technology reached a point where AI classification on real legal documents was defensible and accurate enough to be the primary review mechanism rather than a QC supplement. Cloud storage made per-GB all-in pricing economically viable — the underlying cost of storing one gigabyte for one month on S3 is well under a cent, so the markup that justified $25/GB/month hosted review no longer holds. And no-seat-fee pricing models emerged as platforms recognized that the labor arbitrage was in document classification, not in billing for individual reviewer accounts.
What has not changed: attorney judgment calls on close privilege questions, the need for SOC 2 / HIPAA compliant hosting, and the fundamental workflow of ingestion → review → production. The platform changed; the process shape stayed the same.
The review platform selection is a decision that shapes every downstream cost and workflow choice on a matter. These are the seven criteria that matter most when evaluating platforms in 2026.
DecoverAI was built specifically to eliminate the three biggest friction points that litigation support teams encounter with traditional eDiscovery platforms: slow onboarding, high per-reviewer costs, and privilege log labor.
Onboarding in hours, not days. There is no workspace configuration, no field mapping exercise, no admin certification required. Upload processed data or connect a cloud drive source, configure the review parameters, and invite attorneys — review begins the same day. For matters where speed from collection to first review matters — PI litigation, regulatory inquiries with tight response windows, investigations where the scope is still expanding — this eliminates a common source of delay that traditional platforms build into every engagement.
Multi-model AI classification. Rather than training a single model on a seed set — the Relativity Active Learning approach — DecoverAI runs three or more commercial LLMs simultaneously against the full document set and reconciles their responsiveness, privilege, and PII classifications. This architecture is specifically designed to push recall above 95% on responsive documents — the threshold that matters for defensibility in most federal productions. Documents where models disagree are surfaced for attorney review first; documents where all models agree with high confidence are resolved without attorney intervention, keeping the attorney review queue focused on the genuinely ambiguous population.
Automated privilege log generation. Every document coded as privileged automatically contributes its metadata to a privilege log entry. The AI drafts the privilege basis description — drawing on the communication content, the parties to the communication, and any applicable privilege indicators — and the attorney reviews, edits, and approves each entry before the log is exported. The final privilege log is ready for opposing counsel within minutes of coding completion. Privilege log preparation, which used to consume 20–40 hours of paralegal time on a typical matter, is effectively eliminated as a discrete work item.
No seat fees. At $60/GB/month all-in, there are no per-user charges. Add the client’s in-house team, multiple outside counsel firms, contract reviewers, and a consulting expert — the bill does not change. For litigation support vendors managing multiple concurrent matters, this pricing model dramatically simplifies cost projections and client billing: the only variable is the gigabyte count, which is transparent and verifiable.
SOC 2 Type II and HIPAA certified. DecoverAI maintains current SOC 2 Type II certification and HIPAA compliance for healthcare and medical device matters. The trust center at trust.decover.ai provides current certification status and the security documentation required for corporate and law firm information security reviews, available without an NDA.
For litigation support teams evaluating their platform choices in 2026, the question is not whether AI belongs in the review workflow — it does, definitively, and the defensibility argument has flipped in its direction. The question is whether the AI is a bolt-on add-on to a legacy per-user platform or the native architecture the platform was built around from the beginning. DecoverAI was built as an AI-native review platform, and that architectural decision shows up in every step of the workflow: from same-day onboarding to automated privilege logs to a pricing model that does not penalize teams for getting the right people into the review.