In any work I do, I always start with the “why” before the “how”: what I want to achieve and why the work matters. Only then do I consider the capabilities needed to make it happen. Here, I bring that same lens to Hong Kong’s Plan, examining how a government connects AI adoption with economic direction and institutional capabilities.

Alongside this economy-wide perspective, I draw comparison with a civil engineering project I am working on, where AI agents help prepare site-visit reports for banks, insurance companies and other parties. The two cases differ in scale, but together they help illustrate how strategic intent translates into the practical conditions for AI adoption.

The Plan

The Hong Kong government’s First Five-Year Plan for Economic and Social Development (2026–2030) sets a medium-term direction for the city. It defines development priorities and guides resource allocation, giving society and investors a clearer basis for their expectations. Crucially, it also connects those long-term priorities with annual work plans, budgets, measurement and accountability.

The challenge is not simply to promote a frontier technology such as AI, but to organise the domestic conditions that make it useful. Hong Kong’s Plan approaches these tasks together, connecting AI adoption with economic priorities, capabilities, people and public administration. Read this way, it offers a leadership lesson: choosing applications is only part of the task; the processes and responsibilities around them must also change.

An economic argument, not a technological fashion

Hong Kong’s Plan seeks to strengthen its established advantages in finance, maritime services, trade, aviation and education, and its gateway role between mainland China and the world, while building an international innovation and technology centre around AI. Its AI+ initiative pairs “industries for AI” – capabilities supporting AI itself – with “AI for industries” – adoption across other sectors. The aim is both to improve existing industries and create new value, avoiding AI as merely a cost-cutting tool or an industry disconnected from demand.

The purpose is social as well as economic. With well-being as its ultimate objective, the Plan addresses ageing and declining birth rates within a broader programme for education, healthcare, social protection and public administration. The question is not how much AI is used, but what value it serves: purpose before technology.

In the live case, the firm likewise defined the purpose first: reduce repetitive, template-based report writing and improve consistency and quality for banks and other recipients. At both scales, the tool serves the purpose, not the other way round.

Where the bets are being placed

The Plan does not rank AI industries; it describes connected priorities. AI sits alongside microelectronics, life and health technology, advanced manufacturing and future industries such as embodied intelligence – intelligence built into machines that interact with the physical world. Quantum technology is also named as a future industry, though it is not an AI application.

Health and medical innovation receives substantial attention, with AI presented as a means rather than an end: AI-supported evaluation of drugs and medical devices and upgraded clinical-trial infrastructure put social benefit ahead of technological advancement for its own sake. Shipping follows the same logic: a shift from volume to value through digitalisation, smarter operations and greener activity, supported by AI.

These initiatives are not all AI projects. Together, they describe the commercial setting in which AI is intended to contribute, linking new capabilities with established economic functions. Research, finance, healthcare, manufacturing and logistics can become customers, suppliers and sources of expertise for one another.

In the live case, priorities followed the dependencies between data collection, processing, evaluation and reporting. Hong Kong’s Plan connects industries; the firm connects stages within a workflow. These are not equivalent decisions, but both illustrate why strategy must address how the parts support one another, rather than treating each investment in isolation.

The appendix summarises selected initiatives from the Plan.

Building the route from research to use

The Plan proposes three major innovation and technology parks and five key research and development institutions, including dedicated ones for AI, microelectronics, and life and health technology. More importantly, it connects stages that are usually separate – research, translation into practical outcomes, pilot activity and mass production – and extends that logic across the Greater Bay Area, linking Hong Kong, Macao and neighbouring Guangdong through shared computing infrastructure and easier movement of people, capital and data. The strategic point is not simply to accumulate technology or data, but to connect them with the institutions and activities that can put them to use.

In the live case, the enabling condition was reliable digital site information in formats the agent could use. The scale differs from Hong Kong’s proposed infrastructure, but the principle is similar: technology needs usable data and supporting capabilities.

People are part of the infrastructure

Hong Kong’s Plan develops education, technology and talent together, linking the ability to nurture and attract skilled people with its industrial priorities. Its university-town model connects campuses, industry and residential communities, while AI integration in education and vocational learning is paired with a “Technology for Good” orientation. The strategic lesson is simple: people and skills are part of the infrastructure, not an afterthought.

In the live case, engineers changed how they recorded site information, while the lead engineer moved from drafting reports to evaluating them. This illustrates the people dimension of adoption: roles and working habits must evolve alongside the technology.

Government must change its own operations

The government is not limiting itself to encouraging others. The Plan calls for more digital and intelligent public administration – integrated processes, digital-first services and digital identities for individuals and businesses – and explores an “AI City Brain” that connects departmental data for risk alerts, prediction and coordinated responses.

Social policy follows the same logic, moving from reactive assistance to proactively identifying needs, including through a possible welfare information-sharing platform. Not every measure is AI, but the direction is clear: information should improve the timing and targeting of action. By changing its own operations, the government leads by example rather than by instruction.

In the live case, a revised site-visit checklist aligned information gathering with the agent’s needs. This illustrates the operational change that Hong Kong’s Plan also envisages: adapting existing processes rather than simply adding technology to them.

What we can learn from this

The most transferable part of the Plan may be its least technological: a chain connecting the five-year framework to annual execution. Each year, the Chief Executive’s Policy Address will report on budget and implementation, which are then monitored, assessed and adjusted. This gives the Plan an intended management rhythm, rather than leaving it as a catalogue of ambitions. Its credibility will depend on whether reviews lead to practical adjustments.

I read this as a useful example of strategic governance towards an emerging, transformative technology. The Plan backs its intent with measurable commitments – 105 indicators across economic and social policy, and an ambition to raise research and development spending from 1.63% of GDP in 2024 to 3% after 2030 – and explicitly pairs an “effective market” with a “capable government”.

The shared lesson is that AI adoption depends on decisions beyond the technology itself. Leadership must connect the intended value with the data, capabilities, working practices and responsibilities needed to deliver it. That is what turns a collection of applications into a strategic approach.

But none of this guarantees success. A plan coordinated from the top carries its own risks: priorities can harden before the evidence is in, and execution will ultimately matter more than intent. Data-sharing ambitions such as the “AI City Brain” and cross-border data flows also raise legitimate questions about privacy and oversight – a reminder that governing AI means governing how data is used and protected, not just how technology is adopted.

These risks do not make this roadmap unviable; they make adaptive implementation essential. I read the Five-Year Plan as a public reference for how the government intends to guide change, not as a script to follow without question. The transferable lesson is to establish direction while keeping implementation open to scrutiny and course correction.

For readers looking for a simple takeaway, the Plan and the live case point to five principles for moving from direction to action:

• Start with purpose. Decide what economic and social value AI should serve – strengthening what you already do well and creating new sources of value – before choosing any technology. Live case: faster, standardised reports.
• Choose connected priorities. Place your bets where existing strengths and new capabilities reinforce each other, rather than spreading effort across isolated experiments. Live case: collection, processing, evaluation, reporting.
• Build the enabling conditions. Connect research, talent, data and infrastructure so that ideas can move from pilot to everyday use. Live case: digital site data the agent can read.
• Redesign your own operations. Adapt processes and responsibilities so AI can contribute to the work, rather than simply adding it to existing routines. Live case: a visit checklist aligned with the agent’s information needs.
• Run it on a regular rhythm of accountability. Translate the long-term plan into concrete work plans and budgets, then measure, review and adjust regularly. Live case: the lead engineer signs off every report and tunes the agent when necessary.

The live case reports three practical benefits:

• Time: about four hours to under 30 minutes per report.
• Consistency: reports follow a common standard.
• Role: the lead engineer shifts from drafting to evaluation and sign-off.

Hong Kong’s Plan offers a useful example of leadership organising for change, rather than merely encouraging the use of a new tool. The engineering case illustrates that work at a different scale. A roadmap sets the direction; leadership connects the conditions for action, and accountability keeps implementation responsive as problems emerge. That is the strategic lesson worth taking away.

Appendix: Selected initiatives in Hong Kong’s Five-Year Plan

References:

• The Hong Kong Special Administrative Region of the People's Republic of China, The First Five-Year Plan for Economic and Social Development (2026-2030).
• TNGlobal, “Hong Kong puts AI governance and fintech in first five-year plan”.