Troutman Pepper Leader on Launching AI Assistant Athena

Will Gaus, Troutman Pepper’s Chief Innovation Officer, was interviewed in the September 1, 2023 Law360 article, “Troutman Pepper Leader on Launching AI Assistant Athena.”

William Gaus, Troutman Pepper’s chief innovation officer, told Law360 Pulse Thursday that Athena has all the capabilities of a standard ChatGPT application, including the ability to answer questions and produce written content like articles and blog posts.

The assistant is named after the Greek goddess of wisdom and war, Gaus said. In Greek mythology, Athena is the daughter of Zeus and aided many heroes like Heracles.

“We liked the yin and yang of [the name], the wisdom or the knowledge portion of it balanced with the physicality of conflict and war, and we viewed that being a really good balance in terms of the human element and the physicality of it, balancing with the ChatGPT portion of it too,” he said.

The firm started conceptualizing its digital assistant in June 2021 and kept it as an ongoing project until generative technology became more mature, Gaus said.

…

Why did the firm decide to create its own AI assistant?

This technology is so new for so many of us. We felt that if we could make it available to as many of our firm users as possible in a secure way that complied with all the ethics and security obligations we have to all of our clients, we were going to learn more in the short term of how this could help the firm. Then, map out a much more informed strategic plan or strategic road map of how generative AI can help the firm. And that just goes to how we approach innovation. It’s a collaborative endeavor, and the more voices from around the firm that get to speak up and explore the technology, the better experience we’re going to be able to provide down the road.

What was the most challenging part of this project?

Balancing the reality of what an application like this could do for us with what we know we can do with it eventually. So it’s knowing how disruptive and how much value these types of large language models can have for the firm. All of those things are amazing, and we’ll get there. Putting a fine point on it for what it can do for us today, how can it make us more efficient? Which ultimately I think has led to the adoption of it by people across the firm.

Did the firm have to clean up its data to plug it into Athena, and if so, how did the firm go about doing that?

We are cleaning the data up because we are going at very specific targets of data. We’re making sure that we are doing this in a practical sense, a secure sense. So we’re not opening it up to everything, which then there’d be a ton of data cleanup. We’re going after the areas that we know have been curated, have been cleaned up and are target-rich environments. With any technology solution that leverages data, the data hygiene question comes into it. So we’re approaching this just like any other technology project that relies on data. It has to be clean data if we’re going to get good output.

How did the firm decide that the AI assistant was safe to use for legal work?

First we had the basics down. Athena is located in a secure place as signed off by firm security, as blessed by the general counsel’s office for use. We put a firm policy in place around how attorneys can use this if it is for client use. We have a very proactive monitoring policy in place to make sure that we’re using it in the right way and in accordance with our policy. And we are having a lot of conversations with our clients around their expectation of where and how we use generative AI.

Did you find that your clients are OK with the firm using generative AI with their data?

What we’ve seen is there’s an acknowledgment that generative AI is a capability that every company is going to explore. Our clients, who are taking an interest in this, want to be informed about how we’re using it, how we intend to use it, and if we intend to use it on any of their work. Some want to know that we are doing it or we need to seek their permission to do it. All of it is a very practical framework for making sure that the law firm and client gets this all right. Clients have been open to it, practical, and again, it’s about the dialogue with our clients.

What are the firm’s plans for its digital assistant?

We are going to continue the push for education around how we use Athena, how it can benefit a partner, an associate, someone in marketing, someone in HR, how it can benefit someone every day today to get the most use out of it, and get all the users in the firm prepared for the future of having an assistant, whether it’s a firm-built or a vendor-built assistant, having them build those skills so they’re ready for the future.

We’re doing beta testing of using Athena with our own data for practices, for administrative reasons and things like that. That’s the holy grail of all of these things: using large language models on our own content in a secure way.