From answer machine to problem clarification
Most people ask AI what to do without describing their situation. The book argues for supplying context and contradictions first, letting AI clarify the problem before any solution is discussed.
Founder's book
The book: you have been using AI the wrong way
Written by Harry Lui from the vantage point of everyday Hong Kong business operations, the book explains why most companies stop at asking AI for answers, and how to start instead from business context, real data and a reusable system.
View the Kindle edition on AmazonRegister for a preview chapter
Kindle Edition · Traditional Chinese · This page introduces the Traditional Chinese Kindle edition published on Amazon. The full text is not available for download on this site.
The central thesis
The value of AI is not whether it can produce a presentable answer. It is whether you have given it your company's real situation, data and experience to work on. Treat AI as an answer machine and you get generic answers. Give it context first, then let it find patterns in your own records, and it starts showing why customers do not buy, which failures repeat, and which working methods deserve to become company practice.
Who it is for
- Owners and executives who have tried AI tools but find the gain stops at drafting copy faster.
- Decision-makers who must explain AI investment and return logic to a board or shareholders.
- Operations leads sitting on customer conversations, quotations, complaints or repair logs they have never used.
- Founders of one-person companies and small teams who want repeatable workflows without adding headcount.
- IT, operations and HR leaders accountable for internal adoption.
What readers learn
- How to prepare context for a real business problem: the people, what has been tried, the outcome and the desired result.
- Why prompt technique is not the core company asset, while data, conversations and operating records are.
- How to find patterns in records of customers who did not buy, instead of asking for another sales paragraph.
- How repeatedly explained company knowledge becomes a reusable, auditable system.
- How a one-person company assembles a first AI team from roles, data, standards and deliverables.
- Which judgement, accountability and privacy work must stay with people.
Theme preview
These are the book's main themes, not a full table of contents.
Data is the fuel
Messages, call notes, complaints, repair logs, product photos, enrolment enquiries, transactions and logistics records are all usable fuel. The book sets out how to judge which to organise first.
The patterns behind non-purchase
Objections and lost deals are not random. The book proposes analysing enquiry, CRM and sales data by buying stage and objection type.
Turning experience into institutional memory
The same question should not be re-explained from memory each time. The book discusses turning repeated company knowledge into a reusable, auditable system.
The one-person company's AI team
A one-person company should not merely collect tools. The book argues for designing operations around roles, data, standards and deliverables — like a team, not scattered tools.
Human judgement and governance
AI augments judgement, but accountability, review, governance, privacy and evidence remain human. The book treats this as a premise, not a footnote.
About the author
Harry Lui is the founder of Smark Global. He has worked on enterprise systems, digital operations and transformation projects, and delivers AI training and consulting for Hong Kong organisations. The book distils that practice into an approach companies can adopt themselves.
Related articles
Articles extending the book
These articles extend ideas from the book into everyday Hong Kong operating situations.
AI strategy & operating systems·5 min read
Stop Asking AI for Answers: Start with Business Context
Most companies do not lack AI tools. They ask the tool for an answer before they have made the situation clear.
Learn moreAI strategy & operating systems·5 min read
Prompts Are Not the Core: Turn Enterprise Data into AI Fuel
The same AI tool may produce generic copy for one company and reveal lost orders, rework and product opportunities for another. The difference is usually data, not prompts.
Learn moreAI strategy & operating systems·5 min read
Why Customers Do Not Buy: Use AI to Connect CRM, Enquiries and Sales
A company’s most valuable customer evidence may not be its completed orders, but the conversations that nearly became business and then stopped.
Learn moreAI strategy & operating systems·5 min read
The One-Person Company’s First AI Team: Three Practical Workflows
The greatest constraint in a one-person company is not headcount. It is having to remember, decide and restart every kind of work with the same brain.
Learn more