Foundations of Agentic AI: Principles, Paradigms, and Capabilities
Shivakumar R Goniwada · Apress eBooks · 2026
This chapter lays the foundation for understanding how modern AI evolved from classic machine learning into generative AI, LLMs, and now agentic AI. It begins by tracing the history of generative AI and then moves into core ideas that make this new era possible with transformers, large-scale software, hardware stacks, model training, fine-tuning, reinforcement learning, and the lifecycle of building and operating LLM-based systems. The purpose of this chapter is not only to explain the technology but to help you to see how these pieces fit together as one connected architecture. What begins as a model is never just a model; it depends on software frameworks, accelerators, hardware, training pipelines, evaluation loops, monitoring practices, and deployment choices that shape whether the system remains an experiment or becomes usable enterprise capability.