Agentic Artificial Intelligence Harnessing AI Agents to Reinvent Business, Work, and Life (568 Pages) (Pascal Bornet, Jochen Wirtz etc.) (Z-Library)

📄 File Format: PDF

💾 File Size: 5.3 MB

📖 Read Online ⬇️ Download

Registered users can read the full content for free

Register as a Gaohf Library member to read the complete e-book online for free and enjoy a better reading experience.

📄 Page 1

(This page has no text content)

📄 Page 2

AGENTIC ARTIFICIAL INTELLIGENCE Harnessing AI Agents to Reinvent Business, Work, and Life PASCAL BORNET Jochen Wirtz – Thomas H. Davenport David De Cremer – Brian Evergreen Phil Fersht – Rakesh Gohel – Shail Khiyara

📄 Page 3

Key Contributors to Agentic Artificial Intelligence This book is the result of a unique collaboration among some of the brightest minds in agentic AI—a field that is rapidly reshaping technology and business. The contributors to this book come from diverse backgrounds, including AI researchers, business executives, high-level developers, and hands-on consultants who have implemented AI agents across industries worldwide. Their collective expertise, spanning deep technical knowledge, real- world implementation experience, and strategic business insights, has been essential in shaping this book’s depth and vision. Below, the contributors are listed in alphabetical order by last name: • Ian Barkin • Pierre Louis Bouchard • Nicholas Cravino • Dana Daher • Simon Ellis • Andy Fanning • Olivier Gomez • Kieran Gilmurray • Mohsin Khan • Cassie Kozyrkov • Maxim Ioffe • Nandan Mullakara • Arnaud Morvan • Ramnath Natarajan • Jan Oberhauser • Lasse Rindom • Toran Bruce Richards • Sharbs Shaaya • Pooja Sund Each of these individuals has brought unique perspectives, technical depth, and practical expertise to this book, helping to explore not just what AI agents are, but how they are being built, deployed, and scaled in the real world. To all of you—thank you for your invaluable contributions.

📄 Page 4

Published by World Scientific Publishing Co. Pte. Ltd. 5 Toh Tuck Link, Singapore 596224 USA office: 27 Warren Street, Suite 401-402, Hackensack, NJ 07601 UK office: 57 Shelton Street, Covent Garden, London WC2H 9HE National Library Board, Singapore Cataloguing in Publication Data Name(s): Bornet, Pascal. | Wirtz, Jochen, author. Title: Agentic artificial intelligence : harnessing AI agents to reinvent business, work, and life / Pascal Bornet, Jochen Wirtz [and 6 others]. Description: Singapore : World Scientific Publishing Co. Pte. Ltd., [2025] Identifier(s): ISBN 978-981-98-1566-1 (hardcover) | ISBN 978-981-98-1622-4 (paperback) | ISBN 978-981-98-1567-8 (ebook for institutions) | ISBN 978-981-98-1568-5 (ebook for individuals) Subject(s): LCSH: Artificial intelligence. | Artificial intelligence--Industrial applications. | Artificial intelligence--Social aspects. Classification: DDC 006.3--dc23 British Library Cataloguing-in-Publication Data A catalogue record for this book is available from the British Library. Copyright © 2025 by Pascal Bornet and Jochen Wirtz All rights reserved. For any available supplementary material, please visit https://www.worldscientific.com/worldscibooks/10.1142/14380#t=suppl Desk Editor: Geysilla Jean Ortiz Design and layout: Lionel Seow Printed in Singapore

📄 Page 5

“Agents are (…) bringing about the biggest revolution in computing since we went from typing commands to tapping on icons.” — Bill Gates “AI agents will become the primary way we interact with computers in the future.” — Satya Nadella “The age of agentic AI is here.”— Jensen Huang

📄 Page 6

vii CONTENTS PREFACE: A Journey Toward Human Potential .............xiii INTRODUCTION...................................................................1 The Promise (and Limitations) of AI Agents ................20 What You Will Learn from the Book ............................25 Beyond the Book: Your Online Resources ....................30 Key Terminologies for Understanding AI Agents .........31 PART 1: THE RISE OF AI AGENTS .................................33 CHAPTER 1: Beyond ChatGPT: The Next Evolution of AI ......................................................................35 The Birth of Agentic AI: A Convergence of Powers .....35 Agentic AI for Entrepreneurship and Business .............47 The State of AI Agent Adoption in Companies .............54 CHAPTER 2: The Five Levels of AI Agents: From Automation to Autonomy .....................................................61 Breaking Down the AI Agent’s Capabilities .................61 The Complex Reality of AI Agents’ Capabilities ..........67 The Agentic AI Progression Framework .......................69 The Magic of Progressive Autonomy: Understanding AI Agent Levels ....................................77

📄 Page 7

Agentic Artificial Intelligence viii CHAPTER 3: Inside the Mind of an AI Agent ...................83 Key Specificities of AI Agents ......................................83 Inherent Limitations of AI Agents .................................87 When One Is Not Enough: The Power and Practice of Multi-Agent Systems ..................................90 The Agent’s Dilemma: Balancing Creativity with Reliability ............................................................100 CHAPTER 4: Putting AI Agents to the Test .....................109 Digital Hands: When AI Learned to Use Computers .. 110 Our First Steps with a “Computer Use” AI Agent: The Invoice Test ......................................... 112 When AI Meets the Paperclip Challenge .................... 114 The outcome of the experiment ................................... 118 Lessons learned from the experiments ........................121 PART 2: THE THREE KEYSTONES OF AGENTIC AI .. 125 CHAPTER 5: Action: Teaching AI to Do, Not Just Think ............................................................................129 The Detective’s Dilemma ............................................130 Tools as Building Blocks .............................................133 Inside the AI Agent’s Toolkit .......................................139 From Basic to Advanced Tool Usage ..........................144 When Tools Meet Trust ...............................................154 CHAPTER 6: Reasoning: From Fast to Wise ..................165 AI Reasoning: Introducing The Power of Pause .........167 The Power of Many: Multi-Agent Systems in AI Reasoning ...........................................................185 CHAPTER 7: Memory: Building AI That Learns ...........197 Memory is a Foundation of Intelligence .....................198 The Intricate Dance of Short-Term Memory in AI Agents .................................................................206 The Power of Long-Term Memory: Transforming AI from Tools to Partners .....................216

📄 Page 8

Contents ix Designing and Implementing Long-Term Memory in Agentic AI Systems .....................................................219 Adaptation and Learning through Feedback Loops ....229 Best Practices in Managing Memory for Agents.........243 PART 3: ENTREPRENEURSHIP AND PROFESSIONAL GROWTH WITH AI AGENTS .........253 CHAPTER 8: A Practical Guide For Building Successful AI Agents ...........................................................255 Step 1: Finding the Right Agentic Opportunities ........256 Step 2: Defining AI Agents’ Role and Capabilities .....270 Step 3: Designing AI Agents for Success ....................276 Step 4: Implementing Your AI Agents ........................284 CHAPTER 9: From Ideas to Income: Business Models for the Agent Economy ..........................................323 The Birth of Self-Running Businesses: When AI Became an Entrepreneur ........................................323 Emerging Business Models in the Age of Agentic AI ...................................................................329 Building Opportunities in the Agentic AI Economy: The New App Gold Rush ......................343 PART 4: ENTERPRISE TRANSFORMATION THROUGH AGENTIC AI .................................................355 CHAPTER 10: Human–Agent Collaboration: Leadership, Trust, and Change .........................................357 Mastering Work Design and Change Management at Scale ........................................................................357 Leadership in the Age of AI Agents: Building Trust and Collaboration in Hybrid Teams ...................375 The Foundation: Management Vision and Governance ..................................................................389

📄 Page 9

Agentic Artificial Intelligence x CHAPTER 11: Scaling AI Agents: From Vision to Reality ..............................................................................403 The Right Scaling Approach .......................................403 The Automation Experience Advantage: from Level 2 to Level 3 agents ............................................410 Leveraging Generative AI and AI Agents for a Holistic AI Corporate Transformation ......................415 When Agents Go Rogue: Building Essential Safeguards for AI Systems ..........................................421 CHAPTER 12: Case Study and Use Cases of Agents Across Industries ....................................................427 Case Study: Pioneering Enterprise AI Agent Transformation: Pets at Home .....................................428 Agentic Use Cases Across Functions and Industries ..438 PART 5: FUTURE HORIZONS FOR WORK AND SOCIETY ...................................................................441 CHAPTER 13: The New World of Work ..........................443 Work Reimagined: The Symphony of Human and Machine ................................................................443 This Time Is Different: The Dawn of Agentic AI ........453 Reinventing Education in the Age of AI Agents .........456 CHAPTER 14: Society in the Age of Agents ....................463 Reimagining Human Potential in an Agent-Powered World .................................................463 A Framework for Governing the Future of Agentic AI ... 471 CONCLUSION ...................................................................479 The Next Horizon: Emerging Capabilities ..................480 The Urgency of AI Governance: Building Guardrails Before It’s Too Late ...................................482 Reflection and Broader Implications ...........................484 Your Action Plan .........................................................485 The Power of Choice ...................................................488

📄 Page 10

Contents xi More Resources on Agentic AI ...........................................491 About the Authors ...............................................................493 APPENDICES: Practical Resources .................................501 CHAPTER 2 - The Current Offering Landscape through the Lens of the AI Agent Progression Framework .......... 502 CHAPTER 8 - Example of an AI Agent Identity: Our Newsletter Summarization Agent ........................503 CHAPTER 8 - Example of Error Handling Procedures for our Newsletter Project Agents ............508 CHAPTER 8 - Example of Implementation of an Agent Using a Low-Code Platform ........................510 CHAPTER 12 – Use Cases: Enterprise AI Agent Application .......................................................531 CHAPTER 12 – Use Cases: Personal Productivity AI Agent Applications .................................................545 INDEX ..................................................................................551

📄 Page 11

xiii PREFACE: A JOURNEY TOWARD HUMAN POTENTIAL There’s a profound transformation happening in how we work, live, and create value. While many see this as a purely technological revolution, we see something far more meaningful: an opportunity to redefine the relationship between humans and machines in ways that amplify what makes us uniquely human. We are a diverse team of twenty-seven professionals spanning business, academia, programming, and research, united by a shared vision of how technology can serve humanity. Our backgrounds range from implementing enterprise-scale automation systems to pioneering research in artificial intelligence, consulting with Fortune 500 companies, and studying the societal implications of technological change. What brings us together isn’t just our expertise—it’s our shared belief that technology should enhance human potential rather than replace it. Our journey to this book began years ago, though we didn’t know it at the time. Many of us were among the first to implement intelligent automation systems in major organizations worldwide. We pioneered approaches to combine artificial intelligence with robotic process automation (RPA), creating systems that could handle increasingly complex end-to-end business processes.

📄 Page 12

Agentic Artificial Intelligence xiv This work led some of us to co-author Intelligent Automation in 2020,1 which became a global bestseller and helped organizations rethink their approach to digital transformation. We didn’t realize then that we were laying the groundwork for something even more transformative. The intelligent automation systems we built over the past fifteen years—which combine process automation with artificial intelligence to handle structured workflows—have become the foundation for today’s agentic systems. The progression makes perfect sense: before a system can act autonomously (as agents do), it needs to master the basics of executing processes, handling data, and making decisions within defined parameters. These are exactly the capabilities we’ve spent years refining in intelligent automation systems. This foundation gave us a unique advantage when the latest breakthroughs in generative AI opened the door to modern agentic systems. We had already gained experience with many of the fundamental challenges: how to reliably automate complex processes, how to handle exceptions gracefully, how to integrate with existing systems, and most importantly, how to implement these technologies in ways that enhance rather than replace human capabilities. When companies began exploring agentic systems a few years ago, many naturally evolved from their existing intelligent automation platforms, building upon these proven foundations to create more sophisticated, autonomous capabilities. Yet, we approach this topic with humility. Despite our collective experience—or perhaps because of it—we recognize that we’re all still learning. The field is evolving rapidly, and new possibilities emerge almost daily. What makes our contribution unique is not just our technical or business expertise but also our 1 Pascal Bornet, Ian Barkin, and Jochen Wirtz, 2020. “INTELLIGENT AUTOMATION: Learn how to harness Artificial Intelligence to boost business & make our world more human”. https://www.amazon.com/INTELLIGENT- AUTOMATION-Artificial-Intelligence-business/dp/B08KTDVHHQ

📄 Page 13

PREFACE: A Journey Toward Human Potential xv understanding of how to implement these technologies in ways that serve human flourishing. Our goal isn’t just to explain new technology—we want to give people and businesses the tools to build a better world. A world where workers have more meaningful jobs and a better work-life balance, where companies operate more efficiently while delivering exceptional customer experiences. A world where healthcare systems save more lives through smarter care coordination and schools provide personalized, effective learning for every student. A world where communities can solve complex challenges by using resources more intelligently. AI isn’t just about automation—it’s about creating real impact where it matters most. This book is written for leaders, professionals, entrepreneurs, and curious minds who sense the magnitude of the changes ahead and want to understand how to navigate them. So, whether you’re a business executive looking to transform your organization, a professional wondering about the future of your career, or simply someone interested in how technology will reshape our world, we wrote this book for you. We believe we’re at a pivotal moment in history—one where the decisions we make about how to implement and direct these technologies will have far-reaching implications for generations to come. Through these pages, we’ll share what we’ve learned from our successes and failures, the patterns we’ve observed across industries, and the principles we believe will be crucial for thriving in this new era. Let’s embark on this exploration together, guided not just by technological possibility, but by a vision of what technology can help us become. —The Authors March 2025

📄 Page 14

1 INTRODUCTION Are We Missing the Point with Generative AI? Picture this: Your competitor just announced they’re running their entire operation with a team one-fifth the size of yours, yet they’re growing twice as fast. Their secret? They’ve deployed AI agents that autonomously handle everything from customer service to operations, achieving in hours what takes your team weeks. Sounds far-fetched? It’s happening right now. Let us be provocative here. While most businesses are still figuring out how to use ChatGPT for writing emails and creating chatbots, a new breed of organizations is fundamentally reimagining what’s possible with AI. They’re not just automating tasks—they’re creating self-operating businesses that scale effortlessly, adapt continuously, and never sleep. But here’s the paradox that’s holding most organizations back: We’ve built generative AI systems that can think brilliantly but can’t actually do anything. They can analyze complex data in seconds, write compelling presentations, and offer brilliant insights on any topic. Yet they can’t press a button, send an

📄 Page 15

Agentic Artificial Intelligence 2 email, or make a simple reservation. We’ve created a world of brilliant advisors who can’t lift a finger to help. This situation isn’t just inefficient—it’s actively harmful. In boardrooms and offices across industries, we’re witnessing an alarming trend: The more sophisticated AI becomes at thinking and analyzing, the more humans are forced to handle mechanical, repetitive tasks. Knowledge workers now spend up to 60% of their time on “work about work”—copying data between systems, fact-checking AI-generated content, and manually executing what generative AI recommends.2 As David, one of our co-authors, often says: “We’re treating humans like robots and AI like creatives. It’s time to flip the equation.” Through our decades of experience implementing AI solutions in organizations worldwide, we’ve seen this pattern repeat with alarming consistency. Companies invest millions in cutting-edge AI only to find their employees spending more time managing these systems than doing meaningful work. The machines dream while humans grind. How did we end up here? And, more importantly, how do we fix it? The following three stories, drawn from real experiences, illuminate both the promise and the critical limitations of current generative AI systems. They reveal why traditional approaches are failing and point toward a fundamental shift in how we need to think about artificial intelligence—one that could finally bridge the gap between AI’s ability to think and its ability to act. As you read these stories, they will likely resonate with your own experiences with generative AI. More importantly, you’ll begin to understand why the next evolution in artificial intelligence isn’t about making machines smarter—it’s about making them more capable of autonomous action. 2 Asana, 2025. “Why Work About Work Is Bad,” Asana, https://asana.com/ resources/why-work-about-work-is-bad

📄 Page 16

Introduction 3 The Family Vacation: When Machines Dream and Humans Grind The soft glow of Brian’s laptop illuminated his living room as Saturday evening melted into the night. The house was quiet— his kids finally asleep after their usual bedtime negotiations, his wife reading upstairs. The perfect time, he thought, to plan their long-awaited family vacation to Greece. A trip they’d been promising the kids ever since they’d become obsessed with Greek mythology at school. When Brian opened ChatGPT, the clock read 8:37 PM. He sat down at his computer, determined to plan the perfect family vacation to Greece. Armed with the latest AI technology, he felt confident this would be quick and easy. “Show me a two-week itinerary for a family of four in Greece,” he typed into ChatGPT, adding details about his children’s interests in Greek mythology. Within seconds, the AI produced a masterpiece—a perfectly crafted itinerary filled with hidden gems, local experiences, and thoughtful touches tailored to his family: “Day 1-3: Athens. Begin at the Acropolis during early morning hours to avoid crowds. Your children will be captivated by the interactive exhibits at the Acropolis Museum... Lunch at the family- run Taverna Platanos in the charming Plaka district, where the courtyard fills with the scent of jasmine...” The AI’s suggestions were impressive, even accounting for his son’s love of drawing ancient buildings and his daughter’s fascination with mythology. When Brian asked for an hour- by-hour breakdown, the AI obliged with remarkable precision, including optimal photo opportunities and perfectly timed rest breaks. But as the clock ticked past 10 PM, Brian’s amazement turned to frustration. The “charming family-run” hotel? Permanently closed. The “hidden beach”? Impossible to find on any map. The traditional cooking class? Booked solid for six months.

📄 Page 17

Agentic Artificial Intelligence 4 By 11:30 PM, Brian’s desktop resembled a crime scene investigation: dozens of browser tabs, multiple spreadsheets tracking flight options, screenshots of hotel rooms, and PDFs from tour companies. The AI’s beautiful itinerary sat uselessly in a document while Brian did the real work—checking availability, comparing prices, and attempting to turn the AI’s perfect fantasy into bookable reality. “I would have loved to spend my evening imagining the places we’d visit,” Brian reflected later. “Instead, I spent hours doing the tedious logistics that I thought AI was supposed to handle.” His experience crystallizes what so many of us expect from AI versus what we actually get. We want technology to handle the tedious parts—the endless browsing of flight options, the cross-referencing of hotel reviews, and the mind-numbing task of finding availability across dozens of booking systems. Instead, AI has become remarkably good at the enjoyable parts of planning—dreaming up possibilities, suggesting adventures, painting pictures of perfect moments—while leaving humans to handle all the practical details. The irony wasn’t lost on Brian. Here was one of the most advanced AI systems in the world, capable of writing poetry and explaining quantum physics, yet it couldn’t perform the basic task of checking if a hotel was still in business. It could dream up the perfect vacation but couldn’t book a single flight. Brian finally went to bed at 1 AM, having booked nothing. His browser history told the story: 47 different websites visited, dozens of searches, and multiple abandoned shopping carts on various booking platforms. The AI’s perfect itinerary sat in a document on his desktop, beautiful but useless, like a travel magazine from an alternate reality where everything works exactly as imagined. ***

📄 Page 18

Introduction 5 Brian’s experience with vacation planning reflects a pattern we’ve seen repeatedly across industries and applications. We’ve observed the same limitations, whether it is professionals trying to organize complex project timelines, executives coordinating multi-team initiatives, or entrepreneurs attempting to launch new products. In each case, today’s generative AI systems demonstrate both remark- able capabilities and frustrating limitations. Similar to a room filled with brilliant advisors who are unable to implement their own recommendations, these systems shine in the strategic and creative domains: generating strategies, formulating detailed plans, comprehending complex requirements, offering personalized advice, and crafting compelling narratives. Yet they crucially lack the practical capabilities that would make them truly transformative: • Capability to execute actual actions in the real world • Ability to verify and update real-time information • Power to adapt plans when faced with changing conditions • Capacity to maintain consistent action over time to achieve a goal What’s particularly troubling about our current generative AI landscape is a profound irony that few have recognized: AI has evolved to excel at precisely the wrong things. Think about what excites us and makes us uniquely human— creativity, deep connections, and critical thinking. These are the tasks that fuel fulfillment, innovation, and progress.3 Yet, today’s generative AI excels at them. It can craft a brilliant marketing copy, dream up groundbreaking product ideas, and even engage in sophisticated analysis. Meanwhile, humans are increasingly 3 Cambridge International. “Chapter 4: Innovation and Creativity,” Cambridge International, https://www.cambridgeinternational.org/Images/426483-chapter- 4-innovation-and-creativity.pdf

📄 Page 19

Agentic Artificial Intelligence 6 reduced to data entry, follow-ups, and digital housekeeping— the kind of mind-numbing tasks AI should be handling. This role reversal—where humans become “the robots” connecting various systems while AI dreams up possibilities— points to a fundamental misalignment in our approach to artificial intelligence. But as we discovered in the research world, this misalignment between AI’s capabilities and real-world needs could have far more profound consequences… When AI Met Reality: A Cautionary Tale from the Research World The following story is based on actual events. Names and specific details have been changed to protect confidentiality. Dr. Jessica Ying stared at her computer screen in disbelief. In forty-eight hours, she was supposed to present groundbreaking research on climate change’s impact on global food security at the UN Climate Summit. Her findings were expected to influence international policy and billions in agricultural investment. But as she reviewed the draft her research team had prepared, her heart sank. Three weeks earlier, Jessica had received the call every child dreads—her father had suffered a severe stroke. She’d immediately flown to Singapore to be with him in his final days, delegating the research completion to her capable but inexperienced team of postdocs and research assistants. “Use whatever tools you need,” she’d told them during a rushed video call from the hospital. “Just make sure everything is verified and rock-solid. The world will be watching.” Her team had taken that permission and run with it, embracing AI tools to help complete the massive analysis on schedule. Now, back in her office at the Climate Research Institute, Jessica was discovering the cost of that decision. “Show me how you verified these findings,” she asked her lead researcher, Tom, during an emergency late-night meeting.

📄 Page 20

Introduction 7 Tom pulled up multiple generative AI chat windows, each filled with impressive-looking analysis and citations. “The AI analyzed all our data sets,” he explained. “It found patterns we hadn’t even considered.” But as Jessica dug deeper, her professional alarm bells started ringing. The AI had generated compelling narratives about climate impact on crop yields across Africa—but when she checked the cited papers, they didn’t exist. It produced detailed statistics about farmer adaptation strategies in Southeast Asia, but the numbers didn’t match any known studies. “We thought we were being thorough,” Tom admitted. “We had the AI verify its own findings by cross-referencing across multiple conversations. But we’re now realizing each conversation was operating in isolation, sometimes contradicting the others.” If they had presented this research unchecked, it could have misdirected billions in agricultural investment, influenced international food security policies, damaged Jessica’s twenty- year reputation in climate science, and undermined public trust in climate research itself. Jessica glanced at the photo on her desk—her father at her PhD graduation, beaming with pride. He’d taught her the fundamental principle she’d nearly forgotten: in science, confidence means nothing without verification. With the summit looming, Jessica made a difficult decision. She called the organizers and withdrew from the keynote slot. Her team would need weeks to manually verify every data point, cross-reference every source, and rebuild the analysis from the ground up. This high-stakes near-miss highlighted the dangerous gap between generative AI’s apparent capabilities and its actual limitations. While it could generate impressive-looking research content, it lacked the crucial abilities needed for reliable scientific work: fact-checking, maintaining consistency, comparing sources, and building coherent arguments over time.