Lesson 1: Where AI Fits in Your Business

Course: AI for Business Strategy Pathway: Business (Paid) Level: Intermediate Estimated reading time: 8 minutes

Identifying Real Opportunities (Not Hype)

Every week, another headline tells you AI will transform your industry. Some of those claims are legitimate. Many are not. Your job as a decision-maker isn't to adopt AI everywhere — it's to figure out where it genuinely helps and where it's just noise.

This lesson gives you a practical framework for identifying where AI fits in your specific business, so you can invest your time and money where it actually matters.

The Hype Problem

The AI industry has a marketing problem. Vendors describe every product as "AI-powered" whether the underlying technology is genuinely intelligent or just a basic automation with a chatbot bolted on. This makes it harder for business leaders to separate signal from noise.

Here's a useful mental model: AI is most valuable where you have large volumes of repetitive cognitive work. Not physical labour (that's robotics), not creative strategy (that's still very human), but the middle layer — the pattern recognition, classification, summarisation, and prediction tasks that eat up your team's time.

A Wellington-based logistics company doesn't need AI to decide which markets to enter. But it might benefit enormously from AI that optimises delivery routes, predicts maintenance schedules, or flags anomalies in shipping data. The difference between those two applications isn't just scale — it's fit.

Three Categories of AI Application

When evaluating where AI fits, it helps to think in three categories:

1. Efficiency plays — doing existing work faster

These are the lowest-risk, highest-certainty applications. Your team already does this work; AI just does it more quickly. Examples include:

Drafting routine correspondence and reports
Summarising meeting notes and documents
Processing invoices and purchase orders
Triaging customer enquiries before a human responds
Generating first drafts of marketing copy

Most businesses have dozens of these opportunities hiding in plain sight. They won't make headlines, but they free up real hours every week.

2. Insight plays — seeing what humans miss

AI excels at finding patterns in large datasets that humans would take weeks to spot (or never find at all). Examples include:

Analysing customer behaviour to predict churn
Identifying quality issues in manufacturing data before they become defects
Spotting trends across thousands of support tickets
Forecasting demand more accurately than historical averages

These applications require more data maturity and tend to take longer to implement, but the returns can be significant. A Christchurch-based retailer analysing point-of-sale data with AI might discover buying patterns that reshape their entire inventory strategy.

3. Experience plays — creating something new for customers

This is where AI enables offerings that weren't previously possible at your scale. Examples include:

Personalised product recommendations
Real-time language translation for international customers
24/7 intelligent customer support
Dynamic pricing based on market conditions

Experience plays are the most exciting but also the riskiest. They're customer-facing, harder to control, and more likely to go wrong publicly. Start here only after you've built internal confidence with efficiency and insight plays.

The Opportunity Audit

Before you invest in any AI initiative, run a simple audit across your business. For each department or function, ask:

Volume: How much of this work is repetitive? If a task happens fewer than a handful of times per month, AI probably isn't worth the setup cost.

Consistency: Does this work follow predictable patterns? AI handles structured, pattern-based tasks well. Highly variable, judgment-heavy work is a poor fit today.

Data availability: Do you have (or could you collect) the data needed to train or feed an AI system? No data, no AI — it's that simple.

Cost of errors: What happens if the AI gets it wrong? Summarising meeting notes incorrectly is annoying. Misclassifying a medical referral is dangerous. Match the risk tolerance to the application.

Current pain: Is this a genuine bottleneck or frustration for your team? The best AI projects solve problems people actually have, not problems you think they should have.

A New Zealand Perspective

AI adoption in Aotearoa New Zealand is uneven. Large enterprises and government agencies are investing, but many small-to-medium businesses are still in the early stages. This is actually an advantage — you can learn from overseas mistakes without repeating them.

The NZ market also has specific considerations. Our relatively small scale means some AI applications that make sense for a company processing millions of transactions might not justify the investment here. But our strong services economy — accounting, law, consulting, financial advice — is rich with exactly the kind of cognitive work where AI shines.

The Privacy Act 2020 and the emerging AI guidance from the government also shape what's appropriate. Any AI application handling personal information needs to be assessed through a privacy lens from day one, not retrofitted later.

What AI Cannot Do (Yet)

Being clear-eyed about limitations is just as important as spotting opportunities:

AI cannot replace judgment. It can inform decisions with better data, but the decision itself — especially one involving values, trade-offs, or stakeholder impact — remains human.
AI cannot guarantee accuracy. Every AI system produces errors. Your implementation must account for this with appropriate review processes.
AI cannot manage relationships. Client trust, team morale, supplier negotiations — these are deeply human activities. AI can support them (preparing briefs, drafting talking points) but cannot replace them.
AI does not understand your business context. It processes patterns. You provide meaning. That distinction matters more than most vendors will tell you.

Starting Smart

The businesses that get the most from AI aren't the ones that adopt it fastest. They're the ones that adopt it most deliberately. Start with a genuine problem. Confirm the data exists. Run a small pilot. Measure the results honestly. Then decide whether to scale.

That's the approach we'll build on throughout this course.

Key Takeaways

AI is most valuable where you have high-volume, repetitive cognitive work — not everywhere
Think in three categories: efficiency plays, insight plays, and experience plays — and start with efficiency
Run an opportunity audit before investing: check volume, consistency, data availability, error tolerance, and genuine pain
NZ's services-heavy economy has strong AI opportunities, but scale and privacy considerations matter
AI supports human judgment — it does not replace it

Practical Exercise

Your AI Opportunity Map

List the five most time-consuming repetitive tasks in your business (or department). Be specific — "processing invoices" is better than "admin."
For each task, score it 1–5 on the five audit criteria: volume, consistency, data availability, acceptable error rate, and current pain level.
Add up the scores. Tasks scoring 20 or above are your strongest AI candidates.
For your top-scoring task, write a one-paragraph description of what an AI-assisted version of that work would look like. Who does what? Where does the AI help? Where does a human stay in the loop?
Share your top candidate with a colleague and ask: "Does this match your experience of where we waste time?"

This exercise gives you a concrete starting point for the strategy work we'll cover in Lesson 2.

Knowledge Check

1. Which type of AI application is generally lowest risk and best for getting started?

a) Experience plays — creating new customer-facing offerings b) Insight plays — finding patterns in large datasets c) Efficiency plays — doing existing work faster d) Automation plays — replacing entire job functions

Answer: c) Efficiency plays involve work your team already does, carry the least risk, and build internal confidence before tackling more complex applications.

2. What is the most important prerequisite for any AI application?

a) A large IT budget b) Executive sponsorship from the CEO c) Relevant data that the AI system can work with d) A vendor with an established reputation

Answer: c) Without appropriate data, AI systems cannot function. Data availability is the foundational requirement before any other consideration.

3. In the opportunity audit framework, why does "cost of errors" matter?

a) Because AI systems never make mistakes if properly configured b) Because some tasks have consequences that are too serious for any AI error c) Because it determines the subscription tier you need d) Because errors always indicate the AI vendor is substandard

Answer: b) AI systems produce errors. The audit helps you match applications to risk tolerance — low-consequence tasks are safer starting points than high-stakes ones.