From Showcase to Strategy: Lessons So Far in Building Chatbots for Legal Aid скачать в хорошем качестве

From Showcase to Strategy: Lessons So Far in Building Chatbots for Legal Aid

Time: 02:00 PM CT | Topic: AI Advanced (TRACK) Speakers:Sourav Mohan, Lone Star Legal Aid Ashley Oborn, Lone Star Legal Aid Legal aid organizations are exploring the use of AI-powered chatbots—but building one that actually works responsibly, accurately, and securely takes more than just good intentions. This session shares real lessons learned from our development of three chatbots at Lone Star Legal Aid: Juris (internal legal), LSLA Ask (internal admin support), and Navi (client-facing info and referral).Rather than a demo, this is a mid-journey strategy session focused on what we’ve learned so far: prompting techniques, the differences between internal vs. public-facing tools, incorporating memory and retrieval-augmented generation (RAG), and how to build smarter bots without increasing risk. Attendees will leave with actionable takeaways and practical guidance to jumpstart their own chatbot projects. This project is part of an LSC Technology Initiative Grant and has been developed with replicability in mind. In the final part of the session, we’ll also explore what it takes to manage a project like this from within a legal aid organization. From cross-department coordination and internal user testing to milestone tracking and security planning, this segment will offer a practical look at the non-technical work that’s essential to making chatbot development successful—and sustainable.