Want to Leverage AI But Don’t Know Where to Start? Download The Complete Legal AI Action Plan


Getting Started with Legal AI Tools

Contrary to what the recent surge of attention surrounding artificial intelligence would have you believe, AI technology has been quietly working in the background of the tools you’ve been using every day for over a decade now—serving search recommendations on Google, making personalized viewing suggestions on your streaming platforms, and even assisting legal teams with invoice review.

But now, with the advent of generative AI tools like ChatGPT, the technology has been thrust back into the spotlight. And organizations that were previously slow to adopt AI are quickly realizing that they’ll have to secure AI solutions—and fast—if they don’t want to get left behind.

This is particularly true for legal departments. In fact, a recent survey found that the number of legal teams looking to implement a legal AI tool to help control costs has nearly tripled in the past year alone.

But with so much hype surrounding these tools, how can your in-house legal team sift through all the noise and find the legal AI tools that would serve them best? And what’s the easiest place to get started when it comes to AI?

Understanding AI Fundamentals

To better understand what AI tools can offer, you’ll first need to establish a basic understanding of some of the key concepts surrounding the technology.

Let’s cover some of the most common terms that are likely to pop up during your research, so that you can better judge each AI tool’s offerings without getting bogged down by the terminology.

Artificial Intelligence (AI)

AI broadly describes the ability of computer systems to perform tasks commonly associated with intelligent beings like humans.

Whereas regular computing involves a computer acting upon strict rules regarding inputs and outputs, AI allows a computer to learn from the data it receives and dynamically apply those learnings to solve problems.

Reading and interpreting a piece of text, recognizing and responding to speech commands, plotting the best path for reaching a destination, and solving a puzzle are all examples of tasks to which a computer can apply AI.

Machine Learning (ML)

Machine learning involves a computer learning from existing data and deciding how to respond to new, unknown situations.

An example of machine learning in action would be your email system determining which future emails to mark as spam based on the characteristics of the emails you’ve marked as spam in the past.

Supervised Machine Learning

Supervised machine learning is when humans help to train AI algorithms.

There are two key ways humans help machines learn: by pre-labeling the data the AI algorithm starts with so it has examples of how the data should be analyzed, and by stepping in to help the algorithm deal with new situations when it is not sure how to act.

For example, self-driving cars are trained on image data where objects like stop signs and traffic lights have been labeled by humans. This helps the cars’ AI learn to recognize these objects when seen by their cameras.

However, when the car is unsure how to approach a certain situation (for example, when its vision is blocked or it sees a new kind of traffic sign it hasn’t seen before), it can hand control back to the human driver to deal with the situation. The AI then learns from the human’s actions so it knows how to approach similar situations in the future.

Generative AI

Generative AI is a type of artificial intelligence that can create new content—including text, audio, images, video, and more.

Most generative AI tools can understand and respond to human prompts. ChatGPT, for example, can be asked by a user to create a summary of a large piece of text, write a line of code, or draft a report on the key takeaways from a spreadsheet.

Natural Language Processing (NLP)

Natural language processing is a branch of AI that allows computers to understand human communication like speech and text.

Customer service chatbots, which interpret and respond to text-based questions from users, and Brightflag’s invoice review functionality, which reads and categorizes lawyers’ descriptions of work in billing line items, are examples of natural language processing in action.

Optical Character Recognition (OCR)

Optical character recognition is the process that allows text in PDFs and scanned documents to be converted into a format that machines can read.

Brightflag, for example, uses OCR to automatically scan PDF invoices and pull the relevant billing details out for easy analysis. Without OCR, the data from that PDF invoice would not be readable by the software, and therefore could not be used for reporting and analysis.

What is Legal AI?

Legal AI refers to specially trained technology that harnesses the power of artificial intelligence and applies it to legal work. Unlike general AI models, legal AI is built with an in-depth understanding of legal terminology and workflows, which in turn allows it to more accurately assist with legal tasks like legal invoice review, contract review, and e-discovery.

What Does Legal AI Do Best?

We have a foundational understanding of what AI is and how major aspects of its functionality work. Now it’s time to identify the areas where legal AI tools could benefit your in-house team the most.

The unique strengths common to all types of AI include its ability to:

  • Execute repeatable tasks quickly
  • Distill and summarize large volumes of information
  • Analyze datasets to find patterns, detect outliers, and predict outcomes

For the purposes of legal departments, that means the best way of leveraging legal AI tools’ strengths is to apply them to recurring, time-consuming, and/or data-intensive tasks like:

  • Legal invoice review
  • Data analysis
  • Contract review
  • Document review

What Do Legal AI Tools Struggle With?

While legal AI tools excel at analyzing, labeling, and interpreting data, tasks that demand specialized knowledge and a complete understanding of context, as well as certain types of decision-making, are still better left to members of your legal team.

That’s because AI is constrained by the datasets it has access to. While you could, in theory, consult a generative AI tool for analysis of aspects of a legal matter, it would lack the context required to provide independently actionable advice, such as the strategic priorities and risk appetite of your business. In other words, AI falls short in matching the specialized knowledge and contextual understanding of an attorney—backed by their years of education and experience.

While we’re on the topic of consulting genAI tools on sensitive aspects of your legal work, it’s important to consider the data security measures of the AI tools you use. Open-source systems like ChatGPT use the inputs you provide to train their models, so make sure you never share confidential information with these general-purpose AI tools.

Legal AI Functionality in Action

We’ve established a basic understanding of what AI’s core functionalities are, as well as what the technology does (and doesn’t) excel at. Now let’s look at some concrete examples of how legal teams can effectively apply a legal AI tool to their workflows.

Examples of Machine Learning Use Cases for Legal

Machine learning really shines when it comes to the recurring, high-volume tasks that legal teams usually dread the most.

An AI-backed platform like Brightflag, for example, uses machine learning to review legal invoices and correctly apply billing codes to each line item. These billing codes can then be checked against your outside counsel guidelines in the software, and any violations can be flagged for review (and even auto-rejected) without your attorneys needing to lift a finger.

Zillow's Mark Allen explains how AI-backed invoice review has saved his team time and money.

Machine learning can serve a similar purpose when it comes to contract review and redlining. With the help of machine learning, inconsistent contract language or terms that aren’t compliant with regulations can be quickly identified and highlighted during the contract review process. This means attorneys spend less time poring over a document to flag potential issues, and more time strategizing on how those issues can best be resolved.

Examples of Generative AI Use Cases for Legal

As we mentioned earlier, generative AI excels at producing content and quickly completing high-volume, analytical tasks.

Generative AI also holds potential advantages for legal invoice review. With the power of generative AI, hundreds of invoice line items could be condensed into a short paragraph summary of what work was completed, by which timekeepers, over what period of time—and whether the work was compliant with outside counsel guidelines. The tedious work of sifting through billing details manually would disappear.

There are future use cases for GenAI that hold a lot of promise too. For example, let’s say your GC decides that they want an overview of all the work being instructed to the law firm that you spend the most money with. Instead of having to dig through your reporting tool to find this information—setting the right fields and filters and sifting through all the data—generative AI could return the information you’re looking for in seconds with a simple prompt. What’s more, you could add additional prompts to your original query to get generative AI to pull further insights on things like a summary of the matters the firm is working on and resourcing breakdowns. Generative AI would take the time-consuming experience of data analysis and streamline it into a simple series of requests.

Try a Legal AI Tool for Yourself

Knowing how legal AI tools can be used is one thing, but there’s no substitute for actually taking the technology for a spin and seeing firsthand how it can change your legal department’s workflows.

Free AI tools make it easy to run your own experiments—especially when it comes to generative AI. With ChatGPT, for example, it would only take a few minutes and a few prompts to gauge its utility across various practice areas and content formats.

Some great ways to get your hands dirty and experience what generative AI is capable of could include prompting ChatGPT to:

  • Draft an email to a colleague for you, or create an outline for an upcoming presentation
  • Provide an overview of pertinent regulations and case law in a particular area
  • Write a formula to improve how data is presented in a spreadsheet or pivot table

Important note: Never give ChatGPT confidential information. And be sure to take the results you receive with a grain of salt—accuracy is not guaranteed.

Moving beyond free generative AI tools, many AI-backed legal tech solutions are more than happy to provide legal teams with a hands-on look at their product.

Brightflag, for example, offers self-guided interactive demos of their platform. This allows you to experience firsthand what it would be like to apply Brightflag’s AI-backed tools to tasks like legal invoice review and matter management. You can give the interactive demo a try below:

Let’s Get Started

Congratulations. You’ve taken an important first step in your journey to understanding what legal AI tools are capable of—and finding the one that will return the most value for your in-house team.

If you’d like to learn how Brightflag’s modern, AI-powered e-billing and matter management platform can transform your legal operations check out our interactive product tour below, or schedule a personalized demo today.

Michael Dineen, Brightflag's Director of Data Science, smiling in a gray dress shirt and glasses.

Michael Dineen

Director of Data Science

Michael Dineen first joined Brightflag in 2016 as a Data Scientist before working his way up to the role of Director of Data Science. Prior to joining Brightflag, Michael served as a Senior Analytics Consultant with Presidion. He holds a Master of Science (MSc) degree in Business Intelligence and Data Mining from Technological University Dublin, as well as a post-graduate diploma in Software Development.