K&L Gates and Legal Aid of North Carolina use AI to expand access to legal information

Legal Aid of North Carolina was fielding more than 300,000 hotline calls a year with fewer than 200 lawyers to handle them. The organization needed a way to divert some of that volume, so it built an AI chatbot called LIA - the Legal Information Assistant - with pro bono help from K&L Gates. The tool lives on Legal Aid's website and answers questions around the clock in multiple languages, giving users information about their legal rights and suggesting next steps. "Access to justice begins, really, with access to information," said Scheree M. Gilchrist, the chief innovation officer at Legal Aid of North Carolina. "Without information, people really do not know or understand what their legal issues are, what their legal rights are."

A hotline overwhelmed by demand

Legal Aid of North Carolina is a statewide organization with about 400 employees across 23 offices. Even at that scale, demand outpaced capacity. The hotline was receiving over 300,000 calls a year, and the under 200 lawyers couldn't keep up. The opportunity came at the AI and Access to Justice Summit at Stanford Law School in October 2024. Legal Aid walked out of the inaugural event with a connection to K&L Gates, which was looking for a pro bono AI project. The collaboration produced LIA, a legal chatbot that uses retrieval-augmented generation, or RAG, to pull information from external sources and generate responses. The tool is "AI agnostic" - it has toggled between OpenAI's ChatGPT and Anthropic's Claude since launch to maximize accuracy, Gilchrist said.

How LIA works

Users ask LIA everything from simple questions - the address of the nearest Legal Aid office - to detailed legal questions that might require a professional. Common topics include domestic violence protection orders and fighting evictions. Often the questions are vague because people don't know how to ask them. "They might put something as simple as, 'I'm being kicked out. What do I do?'" Gilchrist said. "LIA is able to sort of take that information and align it with the appropriate subject area, and give a response to hopefully guide them in the right direction." LIA is one of the first legal chatbots in the United States, but not the only one. NYU School of Law professor Sateesh Nori developed Roxanne, a bot that helps tenants with housing issues. For legal professionals tracking this space, AI for Legal resources cover similar developments.

K&L Gates' pro bono role

K&L Gates lawyers work on LIA's back end pro bono. They develop prompts across different areas of law, anticipate questions users might ask, and review the chatbot's answers against source material to check accuracy and clarity. Their feedback helps refine the tool. "In a time where we're facing cuts and reductions in our funding, just being able to tap into a resource like the one that K&L Gates has offered us is sort of a lifeline to keeping LIA available to so many communities in need," Gilchrist said. Amy Groff, a partner at K&L Gates' Harrisburg, Pennsylvania office, said the project gives the firm's lawyers a closer look at how commercial AI tools work under the hood. "It's been interesting for us to be able to see what goes into a tool like this," she said. In the first six months, LIA logged about 50,000 user conversations. The data revealed the highest-need areas and surfaced questions the team hadn't anticipated, which feeds new knowledge materials. The collaboration also shows how firms and nonprofits can scale legal access, said Allison Constance, director of pro bono programs at Legal Aid of North Carolina. "What the K&L Gates team is doing is really helping us build a foundation that will help lots of people," Constance said. "Having them work with us on a structural, foundational-related project is extremely impactful."

Why this matters for legal professionals

LIA shows a model law firms can replicate: structured pro bono that builds infrastructure instead of handling individual cases. For lawyers and paralegals, the practical takeaway is that AI tools require constant human review - checking answers, refining prompts, and keeping knowledge bases current. That work is becoming a distinct part of legal practice, and the skills involved are covered in the AI Learning Path for Paralegals. The project also demonstrates how data from user interactions can guide legal aid priorities. Gilchrist said the organization uses conversation data to identify gaps in legal knowledge and develop new materials in response.