How did High Peak Software build a Technology Assisted Review (TAR) tool to digitize legal workflows?
Director of Marketing
Table of Contents
- What is Technology-Assisted Review?
- Why do we think TAR is extremely beneficial for lawyers?
- How does High Peak’s TAR tool work?
- How can you integrate High Peak’s TAR tool in your workflow?
- What kind of limitations do you need to watch out for?
- Do you want to build an impactful product like our TAR tool?
What is Technology-Assisted Review?
Technology-Assisted Review (TAR) refers to using software to classify documents for the legal review process. Reviewing legal documents is a huge undertaking—and a massively time-consuming process.
Every case has hundreds of documents that lawyers and paralegals must review. While manual review is cumbersome, TAR is meant to take care of most of it.
For this reason, it has become a massive tool in the eDiscovery segment. The main intention is to expedite the review process and give lawyers the time they need to understand the case specifics.
There are several other names, such as computer-assisted review, automated document classification, and predictive coding. While most assume that it involves automated review, that’s not true. Instead, it’s done in tandem with a human reviewer who trains these models to fit their needs.
We can trace the concept’s roots back to 2005, but you can find the actual use of these terms in studies published after 2010. While the industry has been averse to technological improvements, TAR has gained popularity given the benefits it offers.
This article will discuss how TAR helps the legal review process and how lawyers can benefit from its use.
Why do we think TAR is extremely beneficial for lawyers?
The main benefit of using TAR is that it cuts down massive amounts of time spent on manual review. Here’s an in-depth look into how that works.
Early access to information
Lawyers can employ artificial intelligence to understand what information they have in hand and what more they need. A few examples include: who to depose, documents required from the opposing party, affirmative defenses, key trial themes, etc.
Previously, they could only identify these issues when they conducted a manual review. Using this information, they can quickly prepare for depositions when running out of time.
Better organization of data
Using Machine Learning (ML) and Deep Learning (DL) algorithms, users can automatically organize their information under different categories.
The system can remove irrelevant information and highlight the most critical snippets—without the manual input of the reviewer.
The organization and data analysis can be done on a rolling basis—making sure no time is wasted. It benefits different cases like regulatory matters, internal investigations, and contract reviews.
Quicker case resolution
Usually, lawyers collect all the documents in the project’s discovery phase. But unless they complete their manual review, they’re not sure which documents hold merit.
Using an e-discovery platform, you can quickly evaluate and decide where to pursue a settlement, or go through discovery, motion practice, and trial. You mitigate any potential risk and reduce costs by avoiding unnecessary trials.
Reduces review costs
It costs a lot of time, effort, and money to bring in legal reviewers to review the case files manually.
In this case, you can automate and classify what needs to be reviewed and reduce the number of billable hours you’re paying for. It cuts down the costs and gives you more accurate results.
These AI models are iterative, which means it’s constantly improving. In the legal space, it’s more commonly known as TAR 2.0, which uses a continuous active learning (CAL) model.
In TAR 1.0, you can do predictive coding based on seed sets. These seed sets were small portions of data used to train the models. If you need to analyze using different parameters, you’d have to redo the whole process—which is not the case with TAR 2.0.
As long as you feed the data, it’ll keep learning, iterating, and executing based on input parameters. Soon, you’ll notice better classification processes and increased efficiency—increasing your ROI over time.
How does High Peak’s TAR tool work?
The main goal of TAR is that it should help with documentation discovery—or e-discovery. Initial studies about its use indicated that 95% of documents are identified accurately as opposed to 51% by human reviewers. It was one of the reasons why the adoption rates skyrocketed since then.
In brief, the process looks like this:
1. Upload the document
Once the legal firm receives the documents, a subject matter expert feeds them into the system. After which, the TAR tool gets to work.
2. Analysis & review
The system analyzes the documents based on the initial input and categorizes them.
For example, our system can classify the document based on the file type (Excel, PPT, PDF, Image, etc.) and document them. If you need additional categories, we can build the product that way.
3. Relevance & importance
Not everything is essential—and such tools can identify that. For this purpose, the tool will rank the document based on relevance and importance—and segregate them for you.
For example, the screenshot below shows how the data is segregated under different tabs. It uses Optical Character Recognition (OCR) technology to read and upload the text in a machine-readable format.
Using Natural Language Processing (NLP), it segregates the data. The SME trains the algorithms first by employing predictive coding—which means they teach what to prioritize.
4. Repeat & iterate
Once steps 1 to 3 are complete, that doesn’t mean the job is done. To increase the model’s accuracy, the expert repeats the process with more documents.
5. Test the model
As the SME keeps feeding more data, they’re training the model to ensure they’re as accurate as possible.
6. Final integration
Once the software achieves the relevant KPIs, they put it into use. It’s important to note that the KPIs look different for various firms—and it’s best to measure metrics that make sense for your firm.
How can you integrate High Peak’s TAR tool in your workflow?
It’s very simple to integrate these tools into your regular review process. All you have to do is key in your input parameters such as:
- Types of files to process
- How to categorize them
- Specific themes to identify
- Types of sentiments
- What information to prioritize
Based on that, whenever you receive any case files, document them under the specific case, and let the system work. In terms of review stages, here’s how it would work:
Understanding case information
Before proceeding with anything, reviewers and lawyers sift through their data. It helps them create a list of keywords, names, and dates to classify and tag the documents.
During the case, they advise clients on how to take the case forward. If new information becomes available, they don’t waste precious time analyzing the raw data but investigate the insights. Using topic clusters like the one shown below, they can start where they want to—and not from the beginning.
Strategizing next steps
The system stores and classifies the documents. Based on that, they can advise clients on whether to take the case ahead or not. They can also identify key pointers like who to depose, the kind of defenses to use, to take an aggressive/ milder approach, etc.
For example, in our matter intelligence augmentation system, you can create a whole timeline of events. It helps lawyers understand the case’s narrative—and they can pick what’s necessary to build their case.
Tagging relevant documents
Creating a systematic tagging system is necessary. It helps reviewers quickly tag data based on priority, description, etc., so you can find them later. They can pull out relevant information for ongoing cases that span months.
The screenshot below shows different tags based on confidentiality, review stage, and more.
Humans and machines alike are prone to errors. For this reason, it’s important to have internal processes in place to identify such issues. With heightened capabilities like the ones detailed above, it becomes easier to weed out errors.
What kind of limitations do you need to watch out for?
While we’ve discussed what this technology can do for you, let’s understand its challenges. Some of them include:
Issues with transparency
There are instances when documents go missing or aren’t logged correctly. Such situations only lead to severe consequences down the line. To circumvent this, we recommend using software that creates audit logs.
It’s important to document the step-by-step process of using the technology. If you don’t, opposing counsel can question the integrity of the defense. Using this, you can also track who has access and what they’re doing with the information.
Lack of complete automation
It’s an assumption that you can achieve complete automation of the process—which can feel underwhelming when you use it for the first time. That’s one of the reasons why we call our software a “Matter Intelligence Augmentation System” and not an AI-powered one.
Despite that, this could be seen as a benefit as such processes always need a human touch to make critical decisions—and that’s what TAR helps you do.
Needs internal training
As far as any technology goes, it’s essential to train potential product users. While they might be uncomfortable using a new product at first, once they get the hang of it, they’ll be able to work efficiently.
It’s crucial to plan for this during the implementation and transition phase of the project. When you learn how to use it, you save time avoiding unnecessary troubleshooting later.
Do you want to build an impactful product like our TAR tool?
TAR is a huge boon to the legal industry, with many lawyers relying on it to gain better control over their cases. It also helps legal firms maintain their competitive advantage in the industry—and increase their annual revenue.
“What I like most about 44th Street is that it makes sense out of all the discovery data on day one.”Carman Caruso, Carmen D. Caruso Law
If you’re looking for a product development partner to build you a TAR tool just like we’ve done for 44th Street Technologies, reach out today.