You're the CEO or business leader of a mid-sized Hong Kong company. You know AI is important. You've read the headlines. You've seen your competitors experimenting. But you don't know enough about how AI actually works to make real decisions. You can't tell the difference between what's hype and what's actually doable. You don't understand what your team is talking about when they get technical.
And you're running out of time.
This article is your 90-day playbook. It's not a university course. It's not "learn Python in 30 days." It's exactly what you need to know, in the order you need to know it, so you can move from "AI sounds important" to "I can make confident decisions about where AI goes in my business."
Month 1: Understand What AI Actually Does (and What It Doesn't)
Week 1-2: The Conceptual Foundation
Stop thinking of AI as a single technology. AI is a toolbox. Each tool solves a different problem. Your job this month is to understand the categories:
1. **Generative AI** (ChatGPT, Claude): Takes a prompt (text, image, audio) and generates something new. Useful for: writing, code generation, summarization, creative work, customer service automation. Not useful for: financial auditing, high-precision medical diagnosis, legal decisions without human review.
2. **Predictive AI** (Machine learning): Takes historical data and predicts what comes next. Useful for: demand forecasting, churn prediction, fraud detection, customer segmentation. Requires clean data and is slower to implement than generative AI.
3. **Computer Vision** (Image recognition): Analyzes images and identifies objects, patterns, anomalies. Useful for: quality control, security, inventory management. Lower ROI than generative or predictive AI for most SMEs.
4. **Automaton/Workflow AI**: Connects multiple tools and triggers actions based on rules. Useful for: reducing manual approval steps, automating approvals, coordinating between systems. Highest immediate ROI.
Your goal: By the end of week 2, you should be able to listen to someone describe a business problem and say, "That sounds like a generative AI problem" or "That requires predictive AI" or "We could use automation to solve that." If you can't, you haven't understood it yet.
Week 3: Real Business Examples
Read three case studies of AI implementations in industries similar to yours (retail, SaaS, logistics, professional services — whatever applies). Read them as a business leader, not a technologist. Ask: What was the problem? What AI was used? What changed? How long did it take? What went wrong?
Your goal: Find one example that resonates. "That's basically our problem." You'll reference this example for the next three months.
Week 4: What Actually Limits AI Right Now
AI is good at: automating routine, rule-based tasks; pattern recognition in large datasets; generating text, code, and creative content; making predictions based on historical data. AI is bad at: understanding context that requires deep industry knowledge; making ethical decisions; working with extremely small datasets; keeping secrets (data privacy); explaining why it made a specific decision.
Your goal: Know the limitations so you don't propose projects that are guaranteed to fail (like "use AI to handle complex legal disputes without human review").
Month 1 Milestone: Can you describe the four categories of AI and give a business example of each?**
Month 2: Identify Your Specific Opportunities
Week 5-6: Audit Your Own Business
This is the most valuable work you do in 90 days. Spend two hours talking to each department head about their biggest pain: What takes the most time? What's error-prone? What do you manually approve every day? What would change if you could get an answer in 5 minutes instead of 5 hours?
Don't restrict yourself to "Can AI do this?" That's the wrong question. Ask: "What problem is costing us the most money or time?" Then figure out if AI is the answer.
You should walk away from these conversations with a list of 8-12 concrete pain points. Not vague things like "customer service is slow." Specific: "Customer service reps spend 6 hours per day on routine inquiries. That's 48 hours per week. At HK$200/hour loaded cost, that's HK$9,600 per week."
Week 7: Reality Check Your Opportunities
Now look at your list. Which of these pain points have these characteristics?
— High volume (happens hundreds or thousands of times per day/week)
— Rule-based or pattern-based (humans follow a consistent process)
— Measurable (you can track whether it improved)
— Expensive (costs significant money or time)
Your top 3-5 opportunities should meet all four criteria. These are your pilots.
Week 8: The AI Readiness Check
For each of your top three opportunities, answer these questions honestly:
1. Do you have leadership buy-in to actually make changes if this works? (If the answer is "yes, but..." it's actually no.)
2. Can you measure success? (If you can't define success in numbers, you can't know if it worked.)
3. Are you willing to change your process if AI suggests a better way? (If you'll reject AI recommendations because "that's not how we do things," it won't work.)
4. Do you have realistic data? (If your inventory system is from 1995 and full of errors, predictive AI won't work.)
If all three are yes, you have a real pilot candidate. If any are no, that opportunity needs to wait.
Month 2 Milestone: You have 3-5 concrete, specific opportunities for AI implementation, ranked by ROI. Each has clear success metrics.**
Month 3: Plan Your First Implementation
Week 9-10: Pick Your First Battle
Don't try to solve everything at once. Pick your highest-ROI opportunity that also has the lowest risk. This is usually customer service automation or invoice processing (if your data is clean).
Now research: What solutions exist? Claude-powered chatbots, workflow automation tools, document processing AI, forecasting models — what's available off-the-shelf?
Your goal: Get three proposals from vendors. Don't commit. Just understand: What does this cost? How long does it take to deploy? What do you have to change in your process?
Week 11: Build Your Pilot Plan
Work with your team (or a consultant) to define:
— Exact problem you're solving
— Current state (what happens now, how long it takes, what it costs)
— AI solution (what tool, how it works)
— Success metrics (response time, cost per transaction, accuracy, customer satisfaction)
— Timeline (usually 4-12 weeks from decision to live deployment)
— Budget (usually HK$50,000-150,000 for a mid-sized pilot)
— Rollout plan (pilot with a subset of transactions first, measure results, then scale)
Week 12: Make the Decision
You now have enough information. The question isn't "Is AI ready?" The question is "Is our company ready to change how we work?"
If yes: Commit to a 90-day pilot. Allocate budget. Assign an owner. Set weekly review meetings.
If no: Be honest about why. Is it conviction? Is it capacity? Is it budget? Name the real blocker and decide if you're going to remove it or table this project.
Month 3 Milestone: You have a funded, scheduled pilot with clear metrics and leadership commitment.**
Your 90-Day Accountability Framework
Don't just read this and move on. Create a learning checklist. Every Monday, ask yourself:
— Month 1: Can I explain the four types of AI to someone without technical background?
— Month 2: Can I articulate our top three AI opportunities and why they matter?
— Month 3: Do we have a funded pilot with clear metrics?
If the answer to any is no, spend the next week filling that gap. Don't move forward until you have a real yes.
Common Mistakes Leaders Make (Don't Be These People)
"We'll learn more and then decide." This is procrastination. By month 6, your knowledge will be outdated again. Make a decision on the timeline above. Learning happens by doing.
"Our data is too messy for AI." This is sometimes true, but often it's an excuse. Start with your cleanest dataset. Prove the concept. Fix data quality as you go.
"AI will replace my good employees." True, but that's the point. Your question should be: "Will my competition replace my employees' routine work with AI, forcing those employees into less valuable roles?" If yes, beat them to it.
"We need a perfect AI solution." You don't. You need a solution that's 70% of the way there and willing to be wrong sometimes. Deploy fast, measure, adjust.
"The vendor promised it would work." No vendor can promise that. They can promise to deploy it and support it. Whether it actually delivers value depends on your execution, your data, and your willingness to change processes.
The Accelerated Learning Track
If you have 90 days, follow the plan above. If you have 30 days (because your board is pushing), here's what matters:
— Week 1: Understand the four types of AI (generative, predictive, computer vision, automation).
— Week 2: Audit your business and identify your top 3 pain points.
— Week 3: Research solutions and get proposals.
— Week 4: Make a pilot decision and commit budget.
It's rushed, but it's doable. Most leaders who move fast see results faster.
What Comes Next (After 90 Days)
If your pilot succeeds (and it probably will), you'll move into month 4 with a different question: "How do we scale this? What's our next opportunity?"
By month 6-9, you should have three pilots running simultaneously. By month 12, you should have proven ROI that funds the next wave of AI projects from the savings of the first ones.
This is when AI stops being an experiment and starts being how your business operates.
To test your AI readiness and get a personalized assessment of where your business stands, use UD's AI IQ Test at https://ai-staff.ud.hk/iq-test. Answer 10 quick questions about your business, your team, and your conviction. You'll get a clear picture of whether you're ready to move now or where to focus first.
The Bottom Line
You have everything you need to go from zero to AI-fluent in 90 days. You don't need to understand how the algorithm works. You need to understand what problems AI solves, how to measure whether it worked, and when you're ready to move.
The leaders winning with AI right now aren't the smartest. They're the ones who learned fast, made decisions on timelines (not perfectly), and stayed committed through the first implementation bumps.
You can do this. Ninety days. Three months. One pilot project. One clear commitment.
Start this week.
Email: sales@ud.hk | WhatsApp: (852) 9696 7545