How to Build an Agent is not a book about experimenting with AI. It is a practical playbook for leaders who need AI systems that work in the real world.

Most organisations use generative AI. Few see meaningful business results. This book explains why.

Written for executives, senior managers, consultants, and technical leaders, it shows how to move beyond chatbots and prompts to autonomous, outcome-driven AI systems that plan, act, integrate with enterprise software, and deliver measurable ROI.

You will learn how agents work, how they differ from automation, and how to design them safely, economically, and at scale, with a focus on decisions, architecture, governance, and execution.

This is a guide to building digital workers inside real organisations.

Inside the book, you will learn:

• What agentic AI is, and why most companies get it wrong

• Chatbots vs workflows vs autonomous agents

• How memory, retrieval, and reasoning expand capability

• Designing agents across CRM, ERP, email, and data platforms

• Build vs buy vs hybrid, and how to choose

• Governance with auditability, controls, and human oversight

• Measuring ROI that boards and finance teams respect

• Running a four-week pilot that produces decision-grade evidence

• Scaling from one agent to coordinated multi-agent systems

• Avoiding failure patterns that stall AI projects

Includes architectural patterns, decision frameworks, implementation checklists, and templates: business case, pilot charter, readiness assessment, and role definitions for emerging AI leadership positions.

You do not need to be a software engineer. You do need to be responsible for outcomes.

Tired of AI demos that never reach production? This book gives you a clear, structured path forward.

AI adoption is no longer the challenge. Leading AI is.

How to Build an Agent is the field manual for doing exactly that.