Live Webinar · Aug 10 Gen AI for In-House Counsel: how we post-train an LLM to run Document Review at under 5¢/document. Save Your Seat →
80%
Reduction in review volume with AI classification
Per document — vs. $1–$8 with contract attorneys
<1hr
From upload to production-ready output

"What used to take our associates two weeks of full-time review, DecoverAI completed in forty minutes."

How it works
01
Upload your corpus

Emails, PDFs, Slack exports, Office files. Drag-and-drop or bulk ingest. Indexed and ready for review in minutes.

02
AI classifies every document

Responsiveness, privilege, confidentiality tagged across your entire corpus automatically. Review volume cut by up to 80%.

03
Export production-ready output

Bates-numbered docs, privilege log, redaction overlays. Fully audit-trailed. What used to take weeks, done in under an hour.

04
Attorney oversight

Every classification reviewable. Every redaction overridable. AI does the heavy lifting. You maintain full control.

Three methods AI uses to review legal documents

Not all AI document review tools work the same way. The three dominant approaches differ in how they classify documents, how much attorney input they require to start, and how they improve over time.

Method 01

Predictive Coding / TAR

Technology-Assisted Review trains on attorney decisions — documents marked responsive or non-responsive — and uses that seed set to rank the remaining corpus by predicted relevance. Requires a trained sample to begin; accuracy improves as more documents are reviewed.

Method 02 — DecoverAI

Generative AI Classification

Uses large language models to classify privilege basis, confidentiality level, and responsiveness across every document — without requiring a seed set. DecoverAI applies this approach, delivering first-pass results immediately after upload with no attorney pre-training required.

Method 03

Continuous Active Learning (CAL)

A variant of TAR where the model updates in real time as reviewers tag documents. Each tag improves the model's ranking for subsequent documents, making the system progressively more accurate across a long review.

30K
Documents
3
Days
98%
Cost reduction

30,000 documents reviewed in 3 days. 98% cost reduction.

A leading defense firm used DecoverAI for end-to-end production — privilege logs, Bates stamps, and redactions — saving $147K and 25 days versus a traditional contract attorney review. Read the case study →

How AI document review tools compare

Platform AI Classification No Seed Set Required Privilege Log Included Flat Per-GB Pricing
DecoverAI Generative AI $60/GB/mo
Relativity aiR TAR + LLM Add-on Per-user
DISCO AI Predictive coding Add-on Per-GB + seats
Logikcull TAR Manual Per-GB + seats

Common questions about AI document review

What AI tools are used for legal document review?
AI tools for legal document review include machine learning platforms that automatically classify documents for responsiveness, privilege, and confidentiality. Leading platforms include DecoverAI, Relativity aiR, DISCO AI, and Logikcull. DecoverAI uses generative AI classification — applying large language models across the entire corpus without requiring seed sets — and includes all AI review in a flat $60/GB/month price.
How does AI document review work?
AI document review uses large language models (LLMs) and machine learning classifiers to analyze each document for responsiveness, privilege, and relevance. Systems like DecoverAI apply generative AI classification, running documents through models that identify privilege basis, confidentiality designations, and responsiveness determinations without requiring attorney-labeled seed sets. Borderline documents are flagged for attorney quality control before production.
Is AI document review defensible in court?
Yes. Courts including Da Silva Moore v. Publicis Groupe and Pyrrho Investments v. MWB have held that AI-assisted review can be defensible when properly validated. DecoverAI uses validated AI classifiers with documented precision and recall metrics, with mandatory attorney QC on all borderline calls — satisfying the FRCP Rule 26(g) certification standard. Every classification decision is logged in a full audit trail.
How much does AI document review cost?
AI document review costs significantly less than traditional contract attorney review. DecoverAI charges $60 per GB per month, all-in — covering ingestion, AI classification, privilege log generation, Bates numbering, redaction, and production. A 100 GB matter costs approximately $36,000 versus roughly $460,000 under traditional per-document review models — a 92% reduction. Most matters run 5–20 GB, putting the all-in cost at $300–$1,200.
What is TAR in eDiscovery?
TAR stands for Technology-Assisted Review, also called predictive coding. TAR systems train on attorney decisions — documents marked responsive or non-responsive — and use that training to rank the remaining corpus by predicted relevance. TAR requires a seed set of attorney-reviewed documents to begin. By contrast, generative AI classification (used by DecoverAI) applies LLMs directly without requiring a seed set, delivering results immediately after upload.
How does AI reduce document review time?
AI reduces document review time by automatically classifying the entire corpus for responsiveness, privilege, and confidentiality — tasks that contract attorney teams would perform document-by-document at $1–$8 per document. DecoverAI's generative AI cuts review volume by up to 80%, eliminating the bulk of manual first-pass work. The Tax Credit Investigation benchmark: 30,000 documents reviewed in 3 days, versus 6–8 weeks for a traditional contract attorney team.
Can AI review privileged documents?
AI systems classify documents as potentially privileged and can identify the basis for privilege — attorney-client communication, work product doctrine, common interest, and others — but attorneys must make the final withholding decision. DecoverAI's AI flags all potentially privileged documents and generates a privilege log with basis, author, recipient, and description. Attorneys review and override any AI call before production. No document is withheld without attorney sign-off.
Case Studies
Real firms. Real results.
Pricing
Flat rate. No surprises.

One price covers everything. Scales with data, not headcount. No seat fees, no enterprise tier, no contract. Most teams process 5–20 GB per matter for a $300–$1,200 all-in cost versus weeks of associate hours.

$60
/ GB / month
All-in. No seat fees. No contracts.
AI document classification (responsiveness, privilege, confidentiality)
Automated redaction detection with attorney override
Auto-generated privilege log (XLSX)
Bates numbering per firm conventions
Full audit trail for regulatory productions
SOC 2 compliant encrypted hosting
Book a Demo →
No commitment required

See AI document review in action

Upload a real matter. Watch AI classify every document in minutes. No commitment required.

Book a Demo