Most companies have bought artificial intelligence services. But few have become more intelligent as a result. This is the reason why — and what leaders should do about it.

Most large companies can now point to something that looks like AI: a chatbot on the website, a coding assistant for engineers, a few pilots in customer service or operations. On a presentation slide, this can resemble transformation. In the profit and loss account, it is often much harder to see.

That is because buying AI is not the same as becoming intelligent. Many companies have added AI tools, but not changed the way their enterprise actually goes about its business. It is a bit like installing advanced robots in a factory while leaving the production line designed for manual work: The machines may be impressive, but the operating model is still the constraint.

Much of today’s public debate focuses on artificial general intelligence and artificial superintelligence — how intelligent a machine can become as an individual entity. For business leaders, the more pertinent question is different: How intelligent can an enterprise become when it is deliberately designed to use AI well?

That is what artificial enterprise intelligence (AEI) is about. It asks whether an organization is built to convert AI capability into better decisions, stronger execution, and lasting competitive advantage. The same model can produce very different outcomes in different companies. In one, fragmented data, siloed workflows, and weak accountability turn AI into noise and rework. In another, cleaner data, redesigned processes, and stronger leadership discipline allow AI to improve service, productivity, and competitiveness. AI tends to amplify what is already there. If the enterprise is confused, AI can scale confusion. If it is disciplined, AI can scale judgment and execution.

The first wave of enterprise AI has been heavily tool-focused: large language models, copilots, chat interfaces, and personal productivity assistants. These can create genuine benefits. Drafting becomes faster, research takes less time, and some coding tasks speed up. But those gains are not the same as enterprise intelligence. Just as importantly, companies need to develop an enterprise AI portfolio aligned with corporate strategy. Too often, AI initiatives grow as disconnected experiments driven by local enthusiasm rather than enterprise priorities. If a company competes on customer intimacy, its AI investments should strengthen service, personalization, and sales effectiveness. If it competes on operational excellence, the portfolio should focus on planning, automation, supply-chain visibility, and decision speed. AI should be treated not as a collection of tools but as a portfolio of strategic bets designed to reinforce competitive advantage.

Enterprise intelligence emerges when AI is embedded in how work actually gets done — thereby helping the organization sense, analyze, coordinate, decide, and act across workflows that cut across teams and systems. That requires more than a model. It requires redesigned work, trusted data, governance, economic discipline, capable people, and clear leadership accountability.

The six disciplines of AEI

AEI is not a product that can simply be bought and installed. It is an enterprise capability built through a set of linked disciplines that need to work together.

Redesign work for human-machine collaboration: The best performers do not just ask where AI can be inserted into an existing process. They ask how the workflow should be redesigned end-to-end if AI can now analyze, draft, coordinate, and act across multiple stages. That means being explicit about what AI should do autonomously, where human judgment remains essential, when exceptions should escalate and who remains accountable for the result.

Build trustworthy data foundations: No enterprise becomes intelligent on top of poor data. AI makes weak data more dangerous by scaling unreliable outputs and eroding trust in decisions. Serious AI readiness requires a single source of truth for important data, clear ownership, strong quality standards, and robust privacy and security controls.

Govern AI as an enterprise asset: As AI moves into pricing, operations, customer decisions, compliance, and risk, it becomes part of the company’s control environment. Management needs to know what AI systems and agents exist, what data they depend on, what risks they create, who owns them and how decisions can be traced and audited. Without that discipline, AI adoption can accelerate risk as quickly as it accelerates productivity.

The winners over the next few years may not be the companies with the most advanced models. More likely, they will be the ones that build AEI: enterprises with redesigned workflows, trusted data, strong governance, economically disciplined investment, capable people, and an AI portfolio deliberately aligned to strategy. That is the difference between experimenting with AI and using it to build lasting advantage

Manage AI economics, not just AI capability: AI can look affordable in a pilot and become expensive at scale. Costs are often variable, usage-based, and harder to predict than leaders expect. Companies need visibility into where spending is going, which use cases consume the most resources and what business return is being created. AI should not only be technically impressive. It should also be economically governable.

Raise human capability alongside machine capability: Human capability must rise alongside machine capability. AI underperforms when employees are handed tools without the skills, standards, and confidence to use them well. Different groups need different forms of literacy: Directors, executives, frontline teams, and technical specialists do not all require the same training. Without that investment, adoption stalls, value leaks away and governance gaps widen.

Put the board and top team on the hook: Boards and top teams must own the outcome. AI now affects risk, capital allocation, customer outcomes, cyber exposure, and competitiveness, so it belongs squarely in the boardroom. Boards do not need to manage models, but they do need to ask whether management has a coherent enterprise-wide AI strategy or just a growing collection of disconnected experiments. They should press on a few simple questions: Which strategic priorities does the AI portfolio support? How is value measured? Where are the risks? Who is accountable when systems fail or underperform?

Here is a simple test for Hong Kong leaders: For many companies, a few straightforward questions will reveal whether they are building AEI or just collecting pilots. Have more than 10 AI pilots been launched without two workflows being fundamentally redesigned end-to-end? Is AI spending rising without a clear view of cost per decision or business return? Does the board hear about use-case counts, but not value attribution, governance incidents, and strategic priorities? If so, the issue is not technology alone. It is enterprise readiness.

AI is rapidly becoming common. What will remain uncommon — and strategically valuable — is the ability to organize the whole enterprise around it well. The winners over the next few years may not be the companies with the most advanced models. More likely, they will be the ones that build AEI: enterprises with redesigned workflows, trusted data, strong governance, economically disciplined investment, capable people, and an AI portfolio deliberately aligned to strategy. That is the difference between experimenting with AI and using it to build lasting advantage.

The author is former president of the Hong Kong Computer Society and chairman of the Applied Science and Technology Research Institute.

The views do not necessarily reflect those of China Daily.