The Agentic Paradigm

Agentic AI has moved from research demos to the core of how modern software is built. Powered by language models, autonomous agents now reason, remember, use tools, and collaborate to solve problems no single model could—reshaping how applications are designed across every industry.

The Agentic Paradigm is a practitioner's guide to building multi-agent systems end to end. It begins with the anatomy of an agent—perception, memory, reasoning, and action, and shows how to build one using LangChain and LangGraph. It then covers multi-agent communication, coordination patterns, and emerging interoperability protocols, before applying these ideas to real use cases across finance, manufacturing, and healthcare. The closing chapters address what separates a demo from production: testing, observability, security, and building responsible systems aligned with ethical, social, and legal standards.

By the end, you will be able to design, architect, optimize, secure, and observe multi-agent systems and judge confidently when an agentic design is the right choice and when a classical one will serve better.

WHAT YOU WILL LEARN
● Build agents with LangChain and LangGraph.
● Understand perception, memory, reasoning, and action into agents.
● Design coordination in multi-agent systems with proven patterns and protocols.
● Apply agents to finance, manufacturing, and healthcare problems.
● Test, observe, secure, and govern agents for production.

WHO THIS BOOK IS FOR
This book is for software engineers, architects, and technical leads who want to build enterprise-grade agentic applications, as well as data scientists, ML practitioners, and researchers moving into agent development. A basic familiarity with Python and software design can be helpful for students, but no prior experience with agents is assumed.

1. Introduction to Agentic System
2. Understanding a LangChain Agent
3. Memory in Agents
4. Perception in Agents
5. Reasoning and Action in Agents
6. Multi-agent System Communication and Coordination
7. Emerging Protocols
8. Coordination and Communication Patterns in MAS
9. Agents in the Finance Domain
10. Agents in Manufacturing
11. Agents in Healthcare
12. Testing of Agents
13. Observability in MAS
14. Making MAS Secure
15. Incorporating Responsible AI Practices in Agents

Dr. Alpana Dubey is a technology leader, researcher, and innovator with over two decades of experience in industrial research labs. She is Innovation Research Principal Director at Accenture Labs, where she leads research and development in multi-agent systems and physical AI. She has worked across diverse areas spanning software engineering, human–computer interaction, and artificial intelligence. Her work has consistently anticipated major shifts in computing. She articulated a vision for human–AI teaming and built early prototypes of cooperating AI assistants around 2019–2020, well before AI-based agentic systems became mainstream, and went on to pioneer the use of foundation models for robotics.

A Senior Member of IEEE and Vice Chair of the ACM Women India Council, she also serves on the board of directors of iHUB Divyasampark (IIT Roorkee). She has authored over 50 peer- reviewed papers at prestigious conferences and journals and filed more than 55 patents. Her ideas have been featured in Harvard Business Review and as an MIT Sloan impact story. She is a dedicated advocate for women in technology.