Beyond the Chatbot: AI Agents for Defense Firms, Claims Teams, and Legal Departments — Governing AI That Acts, Not Just Answers

Frank Ramos, Partner | Goldberg Segalla

Frank Ramos is a Partner in the Miami office of Goldberg Segalla, where he has defended clients in civil litigation for more than 26 years across retail, product liability, premises liability, trucking, insurance, and commercial disputes. He regularly writes and presents on helping clients adopt artificial intelligence safely, effectively, and lawfully. He is especially well suited to this program: he has written on AI agents in law firms and legal operations for Bloomberg Law, served as Vice Chair of the FDCC AI Committee, and has presented on using AI in claims without creating bad-faith exposure, bringing the perspective of a defense trial lawyer who guides both law firms and insurers through AI adoption.

• Education & Credentials

Mr. Ramos earned his J.D., magna cum laude, from the University of Miami School of Law in 1997 and his B.A., summa cum laude, from Florida International University in 1993. He is admitted in Florida and before the U.S. District Courts for the Middle and Southern Districts of Florida, and holds a Martindale-Hubbell AV Preeminent rating.

• Recognition & Leadership

The Florida Bar named Mr. Ramos its 2025 Mentor of the Year, and Best Lawyers in America has named him Miami’s “Lawyer of the Year” in Personal Injury and Product Liability Litigation – Defendants. He has been recognized by Florida Super Lawyers every year since 2013, serves as 2026 Secretary-Treasurer of the Federation of Defense and Corporate Counsel, and is a past President of the Florida Defense Lawyers Association.

• Professional Involvement

Mr. Ramos is a member of DRI’s Artificial Intelligence Task Force. He has written more than 500 articles and 20 books and has given over 400 presentations to lawyers and claims professionals, including “When AI Meets Claims: Navigating Good Faith and Avoiding Legal Risk” at the CLM Alliance Annual Conference.

• Experience

Mr. Ramos helps clients adopt AI safely and lawfully, navigate AI regulation, and litigate AI-related disputes. His recent writing covers running an AI pilot before an organization-wide rollout and using AI to avoid, rather than create, bad faith. As a trial lawyer, he has taken numerous cases and arbitrations to verdict or award, defending retail and hospitality businesses, manufacturers, and distributors in injury, trucking, product liability, and commercial disputes.

SESSION 1 – From Answering to Acting: What an AI Agent Is, and a Step-by-Step Walkthrough of the Agentic Claim Workflow | Frank Ramos | 2:30pm – 3:30pm

This session gives attorneys and claims professionals a precise understanding of agentic AI — the goal, context, tools, decision logic, and oversight that make up every agentic workflow — and the two-way door / one-way door test for deciding how much autonomy an agent should get. It then follows a realistic ‘CaseFlow Agent’ through a new civil complaint: intake, matter confirmation, document classification, deadline extraction, fact mapping, issue screening, the assignment packet, approval routing, task creation, and updating the system of record — with the stop-points and never-assume rules marked at every step.

BREAK | 3:30pm – 3:40pm

SESSION 2 – Permissions, Failure Modes, and Governance: The Control Architecture That Makes Agentic AI Defensible | Frank Ramos | 3:40pm – 4:40pm

This session builds the control architecture around the agent. It covers permissions as the real foundation — least privilege, read/draft/write/send separation, approval gates placed where consequences begin, and materiality thresholds; the nine failure modes that matter in practice, from the confident wrong answer and the silent omission to prompt injection, compounding errors, automation bias, and vendor changes; the ethics and regulatory overlay, including professional responsibility, confidentiality, billing, and the NAIC’s developing AI risk examination framework for claims organizations; and the deployment playbook: narrow pilots, red-teaming, production scorecards, change management, and incident response.