Build foundational AI skills
- step1
Data and privacy
Before you start experimenting with AI tools, know what data you can safely use.
Begin with non‑sensitive examples such as general product details, anonymised content, templates, or process steps. Avoid entering personal or confidential information unless you’re using a secure, business‑grade AI tool.
For more information on data and privacy when using AI tools, visit our page on safe and smart AI practices.
- step2
Effective prompting
A prompt is the instruction you give an AI tool. Before you begin, think about what the tool can realistically do. Some requests may be outside the model’s capability, even if it gives a confident answer.
Choose the right AI tool for the job, as they have different strengths – for example, do you want to draft content, analyse trends in customer feedback, or summarising information?
When you prepare your prompt, be clear about what result you want. Write specific and precise instructions. Include context, tone, audience, format and add any constraints.
Good prompt example:
“Write a 120‑word product description for electricians. Keep the tone friendly and plain‑English. Include a call‑to‑action. Avoid technical jargon.”Weak prompt example:
“Write a blurb about this product.”AI tools can give confident answers even when they’re wrong, so keep your critical thinking switched on and adjust the prompt if needed.
Treat prompting as a back‑and‑forth task. If the first output isn’t right, reply with what to change or keep. Complete a final quality check and ask the tool to list any assumptions and improvements.
The results you get are only as good as the instructions you give the AI tool, so keep your data accurate and well‑structured to improve results.
Once you’re confident with the basics, try using the best version of that AI (model) available to you. Then you can turn on advanced thinking or reasoning modes for complex tasks – these often help produce clearer, more accurate results.
- step3
Reviewing outputs
Once you receive an output, check that it is accurate, complete, and matches your business’s tone and policies – just like reviewing work from a new team member.
It’s a good idea to:
- confirm facts, numbers, names and dates
- make sure the output reflects your services and products
- check references or claims – AI sometimes misinterprets or fabricates sources (hallucinations)
- ensure the tone matches your brand.
Final approvals should always be made by a person.
You can also ask the AI to check its own answer before you do your final review. This technique is called self‑reflection. It can help spot errors, assumptions or missing detail.
Try a follow‑up prompt such as: “Check your answer for accuracy and list any improvements.”
Common AI terms
| Term | Definition |
| AI (Artificial Intelligence) | Technology that uses algorithms to recognise patterns, generate content or complete tasks that normally require human thinking. |
| Large Language Model (LLM) | A type of AI trained on large amounts of text. LLMs power tools like ChatGPT, Claude, Gemini and Copilot. |
| Model | The “brain” of the AI. It’s the part that has been trained by looking at lots of examples so it can recognise patterns and give answers. |
| AI Assistant | A tool that drafts, rewrites, summarises or analyses information. Best for quick tasks like emails, posts, notes and research. |
| AI Agent | A tool that can also take actions on your behalf, such as sending emails, updating records or moving files. Agents are useful but higher risk and need tighter controls. |
| Prompt | The instruction you give an AI tool. It can include context, tone, format, examples and anything you want the tool to avoid. |
| Prompt Injection | Hidden or unsafe instructions that could cause an AI tool – especially an agent – to act in a way you didn’t intend. These can be hidden in emails, documents or webpages. |
| Input | What you give the AI. For example, a question, prompt, document, or data. |
| Output | What the AI gives back. For example, an answer, summary, image, or recommendation |
| Reasoning or Thinking Mode | A setting in some AI tools that takes more time but produces clearer or more reliable results. |
| Connectors, Integrations, Plugins | Software that lets an AI tool access another system, such as email, documents or CRM data. Helpful for automation but comes with extra privacy and security risks. |
| Shadow AI | When staff use unapproved AI tools at work without oversight. This can expose sensitive data or create compliance risks. |
| Human in the Loop (HITL) | A safety step where a person checks and approves important AI outputs – especially anything public‑facing, financial or staff‑related. |
| Hallucination | When an AI tool produces an answer that sounds confident but is wrong or made up. |
Make the most of free training and resources
Build on your foundational skills and stay up to date with sector changes through AI-focussed training and resources. Make the most of free, online options to maximise your knowledge with minimal investment – particularly if you’re in a trial phase.
Here are some free resources to get you started:
- NZBusiness – publishes articles on business, AI and technology.
- Federation of Small Businesses (FSB UK) – provides an online content hub of AI resources.
- AI Forum New Zealand – provides free AI e-learning modules, events and guidance.
- Ministry of Social Development – provides free online AI courses through their education platform Digital Passport.
These are just a few examples– they’re not a full list of all the free resources available.
Invest in your learning
If you’re ready to take the next step in your learning, there are Private Training Establishments (PTE) across New Zealand that offer specialised courses in AI. These often have a cost attached but can be a smart investment, if you want to upskill your business or embed AI in your processes.
Options for further learning and training include local colleges and universities, who regularly run workshops and short courses on AI. Online learning providers, who deliver AI courses that you can attend remotely.
Many organisations offer AI training, so do your research and find what works best for your business.
Leverage new and existing networks
If you’re learning new skills in AI or assessing its relevance to your business – you're not alone. Many small businesses across New Zealand are on the same journey.
Leverage your existing networks or join new ones to learn from your peers. Industry, regional or AI-specific networks are a great way to access training and connect with other businesses.
Reach out to your business networks or join new ones, to see what support and advice they offer.
For example:
- The Regional Business Partner Network (RBPN) offers business support and subsidised training, including workshops on digital tools and AI.
- Industry associations offer events, training and networking opportunities for your sector. For example, Hospitality New Zealand or The Restaurant Association.
- Regional organisations and charitable trusts offer events and workshops to boost economic development. For example, Development West Coast.
- Regional Chambers of Commerce run events, workshops and training for businesses.
- AI Forum New Zealand runs working groups and events for specific topics and industry sectors.
Learn more about