Slide 1 Cover

Hello everyone. I’m truly delighted to have this opportunity to connect with you in Hong Kong today. I’m here to introduce a new book I’ve been working on—Context Intelligence: How the C-I-M Architecture Reshapes Competitive Paradigms in the Age of AI. This is not a technical lecture, nor will I walk through the book chapter by chapter. Instead, I hope to spend about twenty minutes sharing the core thesis, framework, and cases at the heart of the book.

Over the past two years, we’ve all seen AI capabilities advance at breathtaking speed: large language models, Copilots, Agents, automated workflows—they keep emerging. Many companies have started piloting AI. Yet an increasingly obvious problem has emerged: while the tools grow more powerful, the differences between companies haven’t automatically widened. Everyone can access similar models, buy similar tools, and build similar workflows. So where will real competitive advantage shift to? That’s the question this book sets out to answer.

My core conclusion is simple. What’s truly scarce in the AI era isn’t the models themselves—it’s a company’s deep understanding of its customers, business, organization, and decision‑making contexts. I call this capability Context Intelligence. Today, I’d like you to listen with one question in mind: when every company has access to AI, what still belongs uniquely to you, and remains truly hard to copy?

Slide 2 – Today I’ll Only Answer Three Questions

I’ll keep today’s content tight and focus on three questions.

First, Why now? Why must we talk about Context Intelligence right now, rather than continuing to talk about models, tools, or process automation?

Second, What is Context Intelligence? What exactly is it, and how is it different from data, tags, user profiles, or customer journeys?

Third, How C-I-M works – how does the C-I-M architecture proposed in this book explain why enterprise AI succeeds or fails?

Finally, I’ll bring this framework to life with three real‑world cases from the book: Morgan Stanley’s wealth advisor AI, DBS Bank’s industrialisation of AI, and LinkedIn’s understanding of professional context through its Economic Graph and AI recruiting agents. These three cases represent three distinct paths: how a high‑trust relationship industry uses AI to augment expert judgment, how a bank turns AI into a governable and reusable organisational capability, and how a platform company turns a dynamic graph into a moat in the AI era.

So, today we’re not here to prove that “AI matters” – you already know that. What we’re really here to discuss is: after AI becomes infrastructure, what becomes the new unit of competition for companies?

Slide 3 – Why “Context” Must Be Addressed Now

Over the last two years, companies have gone through roughly three stages of using AI.

The first stage was tool dividends – using AI to write copy, handle customer service, write code, retrieve knowledge. Efficiency gains are clear. But these dividends get copied quickly, because tool capabilities themselves aren’t unique to any single company.

The second stage was process dividends. Companies started embedding AI into workflows, chaining steps together so AI could complete more tasks automatically. This goes further than point tools, but it still mainly stays at the level of “how to execute.” As long as the process path is standard enough, competitors can build similar systems quickly.

The third stage is what this book really focuses on: context dividends. That means AI not only performs tasks, but knows what the right task is in the current situation. It doesn’t just answer questions – it understands what this customer, this team, this constraint, this risk, and this point in time actually mean. In other words, competition is shifting from “Can AI do it?” to “Does AI know how to judge, right now?”

That’s why I argue: the more pervasive AI becomes, the more scarce context becomes. Models will become more open, tools cheaper, interfaces thinner. But a company’s understanding of real business situations, its memory of customer relationships, its judgment of complex boundaries – these still require long‑term accumulation.