The recent explosion in generative AI tools like ChatGPT promises to bring unprecedented capabilities in any field, including legal. But how do you actually—safely—harness those capabilities to make your legal processes better, easier, and faster? Not to mention more reliable and secure? Is that possible?
The answer, broadly, is yes. The key is building AI strategically into your processes, which is something you can do with, for example, an AI-powered business process automation platform.
Business process automation connects systems, automates work, and coordinates people, allowing teams to focus on what actually matters. But the road to resolution begins with understanding what you need to solve, what generative AI tools can do, and how to use them.
One great piece of low-hanging fruit to start with? Using AI-powered process automation to automate legal matter intake.
The potential of automating legal matter intake
First, it’s important to understand what sorts of tasks are slowing down your processes. Legal operations professionals know that certain aspects of legal processes are manual, tedious, and time-consuming. Amid the tasks and decision-making that require the full attention, creativity, and expertise of legal experts, there are too many simple and mundane processes that burn Legal Ops’ time and energy.
Essentially, these tasks are all about intake, triage, and coordination, or ITC. Every organization deals with ITC tasks—the slew of unstructured internal emails, chat messages, and one-off requests that every person and team has to manage day-to-day. Triaging is the task of determining how to prioritize and route these inbound requests, and coordination means ensuring that the requests that come in are effectively dealt with. Effective ITC brings about resolution.
For legal teams, this is more precisely defined as matter management.
It’s not merely that matter intake is manual and tedious; a significant related problem is that, faced with unintuitive legal processes, people will work around and outside of your established system. That leads to confusion and a lack of transparency at best, and serious compliance issues at worst.
By one estimate, 67% of employees routinely skip legal policies, and Legal Ops may spend 4-6 hours per day handling unplanned requests.
Tonkean built the LegalWorks solution to automate most of those tasks. LegalWorks is a part of the Tonkean platform that enables you to automate the legal matter lifecycle, from intake to resolution. You can intake legal requests from anywhere, automate simple requests, and coordinate with lawyers when needed.
Manual matter lifecycle management is too often a ludicrous waste of time and an unnecessary risk vector for organizations.
What if you didn’t have to sit there all day, babysitting an email alias that companies treat like a junk drawer? What if you didn’t have to read each message and forward it to the correct lawyer in the organization? What if you didn’t have to reply to the same handful of similar messages every day, replying with the same forms (like NDAs and contracts) over and over again?
The good news is you don’t.
All that legal matter intake falls into two categories: There are simple requests that can be automatically handled, and there are more complicated requests that require the eye of an attorney. With a process automation platform like Tonkean and the power of ChatGPT, you can triage all incoming requests, then automate simpler tasks and escalate more complex ones.
Here’s how to do it with Tonkean:
Start with an empty module, and select Email as your intake source; you can set up the correct email inbox ahead of time, and send a test email to capture the right headers from your email server
Go to Enterprise Components, create a new data source, search for LegalGPT, and enter your API key; if you don’t have one, open an account at OpenAI
In the Solution Access area, choose the solution you’re going to put it in
Next, you’ll want to map out the “when-do” commands in the module. This is a simple structure by which Tonkean organizes everything in a module. In this case, when an email comes in, we want there to be some action. So you can add a ChatGPT-empowered NLP (natural language processing) Action Block to the Email block. Just select the NLP block from the right-side menu.
First, you’ll enter all the fields in the email block that you want to be present—from, email address, subject, body, and so on. Then, on the right-side menu, add whichever fields you want the NLP to look at.
If you like, add a prefilter (with terms like “Confidential” or what have you) to make sure your organization’s sensitive information doesn’t get fed to ChatGPT as part of the LLM’s training data set.
For sensitive or confidential information, you can tell the platform to send the message to certain individuals in the organization who need to review it.
For messages that are not confidential—like requests for an NDA form—add a new LegalGPT module by selecting it from the right-side menu.
Edit the LegalGPT module: Classify the legal discipline, set urgency rate and reason, and so on.
Add Triggers; use a Match Condition, and add more when/do values so that when LegalGPT is done analyzing, send an email (or whatever you need).
Add a new field to the NLP module to tie a given message to the right “area of law” team.
Add “branches” for each area of law in your organization (eg, contract law, consumer, copyright, etc,)—as well as a branch for if no match is found, so the request will automatically get reviewed by a person you stipulate.
The above is an example of what you can do to automate your legal matter intake in Tonkean. Once you’re set up in Tonkean, these steps take just minutes of real time. (You can see it in action here.)
This example involves Tonkean’s LegalGPT module, but there are more that we built for our LegalWorks suite, including:
How Tonkean enables structured data requests via email
One other thing: Much is made of structured versus unstructured data in the world of AI. Structured data is typically fairly “clean,” in that it’s stored in a particular way in an organized database and is in a predefined format. Unstructured data includes data in native formats (like MP3s or Microsoft Word) that is not organized in a relational database. (Think data warehouse versus data lake.) Obviously, structured data is easier for most people to work with, and it makes using AI simpler.
The way Tonkean enables you to craft a template for email forwards (such as when a legal matter request needs to be elevated to a particular lawyer)—defining parameters like urgency, deadlines, area of law, and so on—actually creates a structured data request.
That structure lets you sort and search everything in your Tonkean dashboard. You can actually filter for fields like urgency, deadline, area of law, and so on, to track what’s happening, who needs to follow up on what task, and the status of past or present legal matter requests.
This is one way Tonkean leverages generative AI that’s empowering to your organization without incurring the risks of data leakage or AI hallucinations. And it speaks to how Tonkean enables transparency and auditability.