How to Design an AI Agent: Architectures, Protocols, and Technical Evaluation
Chapter 2 of Agentic AI in Law and Finance focuses on the architectural principles required for agents to function as cognitive work systems in high-stakes domains.
Open Educational Resources
AI for
Law & Finance
An open textbook series for legal and financial professionals navigating AI—from foundational LLM concepts to agentic systems and governance frameworks.
All content is freely available under Creative Commons licensing.
First Book Now Available
Agentic AI in Law and Finance
253 pages · Paperback & Digital
Chapter Releases
Each chapter is available as a standalone paper on SSRN
Governing AI Agents: Risk, Compliance, and Accountability in Law and Finance
Chapter 3 of Agentic AI in Law and Finance proposes a governance framework for scaling oversight requirements to match each system's risk profile in regulated industries.
What is an Agent? A Conceptual Primer and History of Agents and Agentic AI
Chapter 1 of Agentic AI in Law and Finance addresses the need for definitional clarity regarding 'agents' and 'agentic AI' by drawing on nearly a century of scholarship across eight disciplines.
Learn
Visual Explainers
Go beyond definitions with illustrated guides that break down complex concepts. Each article features custom diagrams and practical frameworks you can apply immediately.
Coming soon:
Governance Models
Memory Architecture
RAG Explained
Principal-Agent Problem
About the Project
AI for Law and Finance is a collection of open educational resources designed for legal and financial professionals navigating the rapid emergence of AI. Written by academics and practitioners, it bridges the gap between technical implementation and professional obligation.
The project covers LLM fundamentals, agentic systems, and governance frameworks. Chapters are being released as working papers on SSRN, with print editions available as standalone books. All content is available under a Creative Commons license.
Now Releasing
Agentic AI in Law and Finance
A comprehensive guide to understanding, designing, and governing AI agents for high-stakes professional domains.
- 1. What Is an Agent?
- 2. How to Design an Agent
- 3. How to Govern an Agent
253 pages · Paperback and Hardcover
Coming Soon
LLM Essentials for Law and Finance
From tokens to tool use—everything you need to work effectively with large language models in regulated industries.
- 1. What Am I Working With?
- 2. How Do I Get It to Reason?
- 3. How Do I Get Reliable Output?
- 4. How Do I Handle Real Documents?
- 5. How Do I Improve Systematically?
420 pages · Paperback
Glossary
59 essential terms for understanding AI agents in law and finance
Core Framework
Agent
A system exhibiting the three foundational properties of Goal, Perception, and Action (GPA). An agent pursues objectives, observes its environment, and takes actions to achieve its goals. This represents Level 1 in the three-level hierarchy.
Core Framework
GPA (Goal, Perception, Action)
The three foundational properties that define minimal agency. Goal provides direction, Perception enables environmental awareness, and Action allows the system to effect change. Together, they form the basis for all agentic behavior.
Governance
Human-in-the-Loop (HITL)
A governance model where humans approve each significant agent action before execution. HITL provides maximum oversight but limits throughput and is appropriate for high-stakes, irreversible actions.
Technical Patterns
RAG (Retrieval-Augmented Generation)
A pattern that enhances language model responses by retrieving relevant documents from a knowledge base before generation. RAG improves accuracy and enables grounding in authoritative sources.
AI & Machine Learning
Hallucination
The generation of plausible-sounding but false or fabricated information by an AI system. In legal contexts, this includes invented case citations or nonexistent statutes; in finance, fabricated data or regulations. Hallucination risk requires verification controls and human oversight.
Legal & Economic
Fiduciary Duty
Legal obligations of loyalty and care that agents owe to principals, requiring agents to act in the principal's best interest rather than their own. When AI agents act on behalf of clients or organizations, questions arise about how fiduciary standards apply.
About the Authors
Researchers, practitioners, and educators at the forefront of legal and financial AI
Michael Bommarito
Serial entrepreneur and researcher at the intersection of AI, law, and finance.
Daniel Katz
Legal technology innovator and professor applying AI to transform legal practice.
Jillian Bommarito
AI governance and compliance specialist helping organizations navigate emerging technology.
Open & Collaborative
The book source is available on GitHub under a Creative Commons license. We welcome contributions, corrections, and feedback from the community.