Selecting a Legal AI Provider

So you’ve done your research into the benefits of legal AI, considered which of your in-house team’s greatest pain points AI can solve, and decided to implement an AI tool to save time and money for your team. Now it’s time to take the final step and select your legal AI provider.

Given all the hard work you’ve put into getting to this point, you don’t want to make a misstep now. In addition to an understanding of the level of support, implementation timelines, and degree of customer satisfaction that the software you’re considering offers, there are other key considerations you should be aware of before you fully commit to a legal AI provider.

In this blog we’ll cover all the factors to keep in mind—from how AI models are trained to important questions regarding data security—that will help ensure you choose the legal AI provider that’s the perfect match for your team.

Why You Need AI That’s Tailor-Made for Legal

Before we take a closer look at how to navigate the final steps of the evaluation process, let’s first make one thing clear: Whatever AI provider you choose needs to offer tools that are specialized for legal work.

The issue with general AI models, like ChatGPT or Microsoft Copilot, is that they’ve been trained on generalized data. This means that, when it comes to legal tasks, these tools lack the specialized knowledge and context necessary to produce accurate, reliable outputs.

A general AI model, for example, is not trained on the nuances of legal terminology. If you present a general model with a legal invoice and ask it to classify all the work featured on it, it likely won’t be able to differentiate between a motion to dismiss and a motion to compel. And as a result, its classification of the work outlined in the invoice would be inaccurate. Not to mention the fact that these general models often use the data you feed them to train their open-source models, so giving them sensitive data is not a risk legal departments are willing to take.

A platform like Brightflag, on the other hand, features secure AI that has been built and trained over the course of a decade to classify legal work. This ensures that its legal invoice review functionality is accurate—which in turn allows our patented AI to easily flag line items that are not in compliance with outside counsel guidelines. By leveraging its focused training and data sets, specialized legal AI tools like Brightflag can make a significantly bigger impact than general AI models.

Best Practices for Legal Tech Selection

There are a few best practices when it comes to evaluating legal tech solutions that lay a strong foundation for ultimately making the right selection—regardless of whether the tool you’re considering has an AI component.

These recommendations will ensure you have a strong shortlist of vendors to choose from before diving into the key considerations surrounding their respective AI functionality.

Get Reviews and Recommendations

The best way to predict the performance of a legal tech tool is to see what current users have to say about using it. Feedback from real users can preemptively address questions you may have about the tool, or surface new potential concerns that might be helpful to get clarification on from the vendor.

Some great sources of customer insights include:

• G2 reviews: Check out what users feel about the product on independent review sites like G2. These help you get a glimpse at the day-to-day experience of using the tool, and are a great source for information on how intuitive the software is to use. They’ll also show you what users are saying about the implementation process and ongoing support.
• Legal professionals in your network or on LinkedIn: Considering how quickly AI tools are being adopted by legal teams, there’s a high likelihood someone in your professional network has had experience implementing the tools you’re considering. Reach out to them to see what their experience has been with a particular provider. It doesn’t hurt to learn what they would have done differently during their selection process.
• Message boards (e.g. CLOC): Message boards are a great resource for getting informed opinions from others in your professional community. Here you can ask for recommendations from those who had similar needs to yours when they were onboarding their AI tool, and determine what the community’s consensus is on questions you might have.

Ensure the Functionality Meets Your Needs

Every legal tech tool is different, and focuses on a specific set of needs. That’s why it’s important to choose one that matches your legal department’s unique set of requirements.

Determine what you care about most. Is it saving time? Then you’ll want to choose the tool that leverages AI to empower its automation capabilities. Is it important to ensure easy adoption by end-users? Then make sure you select an AI provider that has built a reputation for the ease of use of its software.

Think About Implementation and Support

A legal tech tool is only useful if you can smoothly and successfully implement it.

Brightflag’s average implementation time is just 45 days. But others can take multiple years—and sometimes never get fully implemented. So make sure you have a clear timeline from your provider on how long the process typically takes. And don’t hesitate to ask customers with similar-sized legal departments in your network what their implementation process was like with the vendor in question.

Additionally, once you’re up and running, you’ll need a trusted guide to help you get the most from your AI tool. That’s where your AI provider’s support model comes into play. A responsive, engaged customer success manager can serve as an extension of your in-house legal team, helping you troubleshoot problems and use an AI tool to its full potential. So be sure to select a provider with a strong reputation for impeccable customer service.

Key Considerations When Choosing a Legal AI Provider

You now have a bird’s eye view of where the legal AI providers you’re considering stand in terms of customer satisfaction and how well they can address your legal team’s needs. But now it’s time to sift through the finer details of their AI offerings and better understand how they differentiate themselves in a few key areas.

Model Training and Configuration

Ideally, the AI you decide to onboard will work right out of the box and won’t need your team to train it. That’s the case with Brightflag.

However, some solutions will require training on your own data. If they do, you’ll want to know how long this training takes and what’s required from your team.

For example, will you have to send over all your contracts so the AI can be trained to find your key terms and clauses? This is good to know, as it allows you to plot out the logistics required to get the AI up and running, and establish a timeline for when you might begin seeing value from your chosen tool.

Accuracy

One thing you’ll want to know about the AI models you’re considering is how accurate they are—and how they go about remedying inaccuracies.

There are certain aspects of your in-house team’s work where complete accuracy from your AI tool is vital (e.g. when drafting court documents). On the other hand, some functions—like aggregate reporting—allow slightly more room for error.

Regardless of the degree of rigor required though, all AI tools need specific legal training to approach an acceptable level of accuracy. They also need to have safeguards in place to ensure they appropriately avoid and/or react to errors.

Supervised machine learning is an example of one such safeguard. If an AI tool employs supervised machine learning, it means that a dedicated human is on hand to effectively “teach” the AI what the correct outputs are until it eventually learns to make the correct decisions on its own. Another popular guardrail is user validation, where an AI assistant will repeat requests back to users to confirm it understands the request correctly before responding.

Privacy

Legal information is incredibly sensitive, so your legal department will want to have a full understanding of where all the information shared with your legal AI tool is stored, and how it is shared.

Typically, it’s best to ensure that your in-house team’s data will be kept on servers within your preferred jurisdiction. For example, if you are a US-headquartered company, you may want your data to reside on servers in the US, while if you’re a European company, you’ll prefer European data residence. And importantly, your data should not be shared with third parties but kept on the servers of your legal AI provider.

This is an area where general models like ChatGPT fall short, because they often pool all inputs that are entered into their system to train their model. This in turn creates potential privacy issues, as we saw in the Samsung source code leak that originated from ChatGPT in 2023.

Data Security

How stringent are the security protocols of the legal AI provider you’re considering?

Leaks and external hacks can be devastating—and can even open up your organization to increased liability. That’s why you want to ensure your legal AI provider is as vigilant as possible when it comes to data security.

The best AI providers will carry internationally recognized security certifications to give your legal team peace of mind. You’ll want to ensure your vendors are, at the very least, SOC 1 Type 2 and SOC 2 Type 2 compliant, and that they have ISO/IEC 27001 certification.

Your Provider’s AI Roadmap

Things move fast in the AI space, so the last thing you want to do is choose an AI vendor that can’t keep pace.

Having a sense of what a provider’s AI roadmap looks like will key you into how innovative and ambitious they are with their product offerings. The best AI providers are the ones that can deliver a world-class tool now while also striving to keep your legal team ahead of the curve by constantly adding new features.

Remember: AI is an investment, so the more a vendor prioritizes staying on the cutting edge, the greater value your legal team will receive from the product over time.

Selecting Your Legal AI Provider

You’ve put a lot of time and effort into getting to the point where you’re ready to select a legal AI provider. Now—armed with an understanding of how to accurately assess their finer qualities—you’re ready to make an informed decision and help advance your legal team’s capabilities.

To learn more about how Brightflag’s modern, AI-backed e-billing and matter management platform ticks all the above boxes and can help transform your legal department, book a demo today.