��è��Ƶ

��DzԳٲ����� �į�į �į�į�į�į�į&�Բ�����;

Maria Duvuvuei    �į�į�į�į�į&�Բ�����;

Communications Strategist      

(202) 295-1542    �į�į�į�į�į&�Բ�����;

media@lsc.gov 

Contact Us      �į�į�į�į�į&�Բ�����;

WASHINGTON – A highlights one Texas nonprofit law firm’s efforts to build artificial intelligence (AI) chatbots. Lone Star Legal Aid (LSLA) developed three different chatbots to improve internal efficiency, centralize information needed by lawyers, and make legal information more easily accessible for the public. 

Host Cat Moon interviewed LSLA’s Ashley Oborn, director of data analytics, and Sourav Mohan, data analyst and lead developer on the , for the new episode of the ��è��Ƶ’s (��è��Ƶ) podcast, Talk Justice.

LSLA has 11 offices serving clients across 76 counties and handles about 25,000 cases each year. When discussion around AI came to the forefront, Oborn considered how this technology might help LSLA staff and attorneys better manage their workloads so they could serve more clients.

“We're always willing to do something and be a certain level of scrappy — get our hands dirty — especially if we're trying to make a difference,” said Oborn.

Oborn wanted to leverage her team’s skills, like Mohan’s passion for Python coding, to develop AI chatbots internally, tailored to LSLA’s specific needs. LSLA received a Technology Initiative Grant from ��è��Ƶ to fund the two-year project in 2024.

The first chatbot they created is called Juris, and it functions as a legal research tool for internal use by LSLA staff. 

“Juris wasn't created to replace other platforms, but mostly [to] curate the way our staff conducts legal research,” said Oborn. “So, gathering up secondary sources and pairing them with black-letter law that's already available on platforms such as Westlaw.”

Another chatbot they developed is LSLAsks, which is also for internal use and serves as an administrative assistant, gathering information on policies, forms and answers to questions frequently asked by staff in one central place. 

“[One] passion of mine, in particular, is centralizing information and getting rid of all these different fragmented sources to create that efficiency and create that productivity for everybody,” said Oborn. 

The third chatbot is externally facing. Named Navi, this chatbot is designed to help users determine if their issue is a legal problem, provide referrals to appropriate resources and share valuable self-help materials for non-legal issues. The chatbot offers information specific to LSLA’s region and puts things in plain language.

Oborn explained they relied on legal experts to ensure that all the information they supplied to the AI chatbot was accurate. Mohan said that through specific prompting techniques, they were able to adapt the chatbot’s behavior to improve its functionality and mitigate risks. 

"I think the main thing to understand about large language models is that it wants to give you an answer no matter what, and a lot of times it wants to agree with you,” Mohan said. “So those were the things we wanted to combat."

Large language models are a type of AI tool trained to work from a supplied dataset. Through prompting, Mohan was able to give the chatbots restraints, making it so that when the AI does not have sufficient information to answer an inquiry, it will clearly state that to the user.

Mohan said the human aspect of legal services remains vital.

“The idea that [AI] can replace a human person in the whole transaction is, I think, a little overblown,” said Mohan. “But at the same time, it can be used to make everyone's life easier so we can serve more people and improve our service area.”

This project is one of several highlighted in ��è��Ƶ's Peer Learning Lab, a space for legal services innovators to share ideas and learn from one another. The first lab, launched in 2025, focuses on the use of generative AI in legal services. Learn more at lsc.gov/learninglab. 

To hear the rest of this conversation, listen to the  online, on Spotify, YouTube or Apple Podcasts. The podcast is sponsored by ��è��Ƶ’s