What Is Technology-Assisted Review (TAR)

The average enterprise litigation matter today involves millions of documents. Emails, Slack threads, Teams chats, cloud files, archived databases, the data doesn't stop accumulating just because a lawsuit has been filed. 

Reviewing all of it manually isn't just expensive; at today's data volumes, it's no longer realistic. That's the problem technology-assisted review (TAR) was built to solve.

TAR is an AI-powered approach to eDiscovery document review that uses machine learning to analyze large datasets, identify likely relevant documents, and surface them to human reviewers first, dramatically reducing time and cost without sacrificing accuracy or defensibility.

TAR in eDiscovery: More Than Just Automation

Technology-assisted review in eDiscovery sits at the heart of the document review phase, the stage that, according to the American Bar Association, accounts for roughly 80% of total litigation costs. 

Technology-assisted review in eDiscovery sits at the heart of the document review phase, the stage that continues to dominate discovery spend. While review accounted for roughly 64% of total eDiscovery costs in 2024, projections suggest it will decline to around 52% by 2029 as AI-driven efficiencies and greater investment in upstream tasks reshape the cost structure.

TAR doesn't replace the document review process; it restructures it. Instead of attorneys linearly working through every document in a collection, TAR ranks and prioritizes the dataset so that legal teams spend their time where it matters most: on the documents most likely to contain responsive, privileged, or strategically significant content.

It's not automation for automation's sake. It's a force multiplier for legal judgment.

How Technology-Assisted Review Works

At its core, TAR document review is a feedback loop between human expertise and machine intelligence. The process typically unfolds in three stages:

Step 1: Seed Set Creation

A subject matter expert, usually a senior attorney who understands the case theory, reviews and codes a representative sample of documents, classifying each as relevant or non-relevant. This sample becomes the "seed set": the training data the machine will learn from.

Step 2: Algorithm Training

The TAR system analyzes the coded seed set and extracts relevance patterns such as vocabulary, context, metadata, and document structure. It then applies those patterns to the broader document collection, assigning each document a relevance score.

Step 3: Document Ranking & Review

Documents are batched and surfaced to reviewers in order of ranked relevance. The highest-scoring documents reach reviewers first. In more advanced implementations, the system continues refining its predictions in real time as reviewers code each new document.

TAR 1.0 vs TAR 2.0: What's the Difference?

TAR technology has evolved significantly since its early adoption, and the distinction between TAR 1.0 and TAR 2.0 is one every legal professional should understand.

TAR 1.0 (Predictive Coding) 

It operates in two distinct phases: first, a structured training phase where senior reviewers code a seed set until the model achieves statistical stability; then, a separate application phase where the trained model scores the entire document collection. 

The training stops before the review begins. This approach works well when parties need to agree on a defined protocol upfront, and the DOJ's Predictive Coding Model Agreement, for example, follows TAR 1.0 conventions.

TAR 2.0 (Continuous Active Learning / CAL) 

It removes the hard line between training and review. The algorithm and human reviewers work in parallel throughout the entire process, every document a reviewer codes teaches the model something new, and the model immediately re-ranks remaining documents based on that learning. 

There's no fixed seed set, no stopping point for training. The system simply gets smarter with each decision, presenting the most likely-relevant documents to reviewers continuously until the collection reaches saturation.

TAR 2.0 has become the dominant approach in large-scale eDiscovery, with cases using CAL workflows seeing a 40–60% reduction in the number of documents requiring human eyes.

Why Technology-Assisted Review Is Now Standard Practice

TAR gained legal legitimacy in 2012 when Federal Magistrate Judge Andrew Peck validated its use in Da Silva Moore v. Publicis Groupe, affirming that computer-assisted review is an acceptable and often preferable approach to eDiscovery. Since then, courts have broadly accepted TAR protocols, provided the process is transparent, documented, and defensible.

The business case is just as compelling as the legal one:

• Speed: TAR can process and rank millions of documents in the time it would take a human team to manually review thousands. 
• Consistency: Algorithmic review applies the same relevance standard across every document, eliminating the variance that comes with large teams of contract reviewers. 
• Cost reduction: By deprioritizing clearly non-responsive documents, TAR dramatically reduces billable review hours. 
• Risk reduction: Audit trails, recall metrics, and validation sampling make TAR workflows highly defensible in court. 

What to Look for in TAR Software

Not all technology-assisted review software is built equally. When evaluating document review technology for your firm or legal team, prioritize platforms that offer:

• CAL/TAR 2.0 support: Continuous active learning as the default, not an add-on. 
• Transparent relevance scoring: Reviewers should be able to see why a document was ranked high or low. 
• Built-in quality control: Random sampling, elusion testing, and conflicting-decision reports. 
• Defensible audit trails: Every coding decision logged and exportable for court or opposing counsel. 
• Scalability: Proven performance on collections in the millions, not just thousands. 
• Integration with broader eDiscovery workflows: TAR should connect seamlessly to your collection, processing, and production stages. 

From TAR to Trial Strategy: All in One Platform

Technology-assisted review has moved from emerging legal technology to industry standard, and for good reason. It's faster than manual review, more consistent than large attorney teams, and now fully court-accepted. As datasets grow and AI capabilities deepen, the distinction between "using TAR" and "winning more efficiently" will only sharpen.

Whether you're navigating a high-volume litigation, a regulatory investigation, or a cross-border data production, the right TAR eDiscovery platform transforms what's possible.

See how Venio Systems brings TAR, CAL, and AI-powered document review into a single, scalable platform. Contact us today to explore how your team can review smarter without reviewing more.

‍