Lone Star Legal Aid builds three AI chatbots to centralize resources and help Texans navigate civil legal issues

Kat Moon, host of the Legal Services Corporation podcast Talk Justice, interviewed Ashley Oborn, director of data analytics at Lone Star Legal Aid, and Saru Mohan, a data analyst and the lead developer on the project, about the nonprofit’s three‑bot AI initiative.

Ashley Oborn said the project began more than three years ago as a small, internal experiment after the rise of generative AI. “Juris is our proof of concept,” Oborn said, describing it as a legal‑research chatbot designed not to replace existing platforms but to curate research results for Lone Star advocates. She said Lone Star has 11 offices serving 76 counties and handled more than 25,000 cases in 2024, which helped motivate the team to seek tools that increase productivity.

Oborn described the other two bots as having distinct roles: Elsewhere is an internal administrative assistant that aggregates policies, forms and cross‑departmental FAQs into a single, searchable workplace knowledge base; Navi is the only external‑facing bot and functions as a plain‑language resource hub for civil legal issues and referrals, with a geographic element that can direct users to local legal‑aid providers or social services.

On sourcing and trust, Oborn said the team prioritized curated inputs. “We wanted to really curate and control the information going into these chatbots so that the information going in wasn’t garbage,” she said, explaining that subject‑matter experts were engaged to map every legal issue and to identify trustworthy source materials before any prompting was applied.

Mohan described technical safeguards used to limit unreliable outputs. “If you cannot answer based on the sources that we give it, then don’t answer at all,” he said, adding that internal chatbots show citations under answers — chapter, section, page and short excerpts — so attorneys can check provenance instead of relying on the bot as a standalone authority.

Oborn said pilots and public demonstration helped secure organizational buy‑in. Debuting Juris at the Innovations and Technology Conference (ITC) provided national visibility and internal momentum, she said, and the project timeline was deliberately stretched to about two years to accommodate team members’ full‑time responsibilities.

Asked about scale, Oborn framed AI as a tool to address a pressing capacity gap. She told the podcast that, in Lone Star’s service area, meeting the full legal need would require one attorney to serve roughly 17,000 clients, and she cited a commonly discussed access‑to‑justice shortfall (quoted in the episode as “92%”). Those figures were presented by Oborn as a rationale for using technology to expand reach, not as independently verified statistics within the episode.

Oborn also told listeners where to learn more: the ITC presentation is available on the Legal Services Corporation’s YouTube channel; the LSC AI peer learning lab hub is at lsc.gov/learninglab; and related case studies are posted at justicebench.org and on Lone Star Legal Aid’s project pages. She said Lone Star would launch its two internal bots to staff the same week Mohan was scheduled to present on the rollout.

The episode ended with the standard LSC disclaimer that guest views are their own and that the podcast content is informational and not a substitute for legal advice.