I’m sure there’s some fancy triangle or quadrant framework out there that lays out the "perfect future-proof AI skill stack". After mucking around with AI for a few years, here's where I'd start if I were just beginning.
I'm pulling together the nuts and bolts of these skills together in a longer, more detailed format, with practical steps. But for now, I wanted to share these thoughts and see what feedback comes of it...
1. AI Fluency
This is the new digital literacy. Everyone throws the term around, but what it really means to me is: “Can you work with AI like it’s a co-worker, not a magic trick?”
Most people think AI Fluency is something to do with Prompt Engineering. This is true to an extent, and it's a component. To me it's more about knowing:
what the AI is good at
how to steer it properly
how to use it across your actual work
how to critique and improve what it gives back
It’s less “what’s the best prompt for LinkedIn content” and more “how would I get these AI tools and assistants to research, draft, rewrite, summarise, and package something so I don’t spend 4 hours on it”.
Tools to Build Fluency:
ChatGPT (GPT-4o) for general use, comms, research, structure
ChatGPT (o3) for problem solving, analysis and reasoning
Claude Sonnet 4 for creative writing
Claude Opus 4.1 for big document breakdowns (or when you want more nuance)
Perplexity for research and source checking
NotebookLM is in a league of its own for organising and reasoning over large sets of information
Custom GPTs and Claude Projects to save recurring tasks and train them on your tone/process
Anthropic’s AI Fluency course is actually solid if you want a framework (Delegation, Description, Discernment, Diligence)
Gemini and ChatGPT’s Deep Research agents are, for me, the most underappreciated way to let AI run wild across the internet and dig up everything on a topic
Wispr Flow is my pick when I want to use and feel AI in a completely different way. The accuracy is unparalleled
Daily Practice:
Improve answers by building on them step by step
Ask the AI to respond as a different person or role
Try tricky questions to see where it struggles
Check answers using another AI or trusted source
Start with a broad question, then narrow it down
Ask for the same answer in list, table, or paragraph form
Take a great answer and figure out what prompt made it
Keep a list of prompts and ideas that worked well
This is about building muscle memory. You get fluent by actually using it.
2. AI Automations and Workflows
If AI Fluency is about working faster, automations are about not needing to work at all for certain stuff.
This is where the time compounding starts. You build something once that saves you time every day.
For me, it’s been a combo of:
pull files from gmail and store them in google drive automatically
generate weekly reports and summaries without manual work
rename and format documents the same way every time
clean up and organise email newsletters
synthesise youtube videos into key points and store them in notes
summarise email newsletters into short, actionable takeaways
create documents and folders automatically from form submissions
sync information between notion, google sheets, and slack
save linkedin and online research into a template with pre-filled interview questions
push meeting notes from slack into a searchable second brain
send reminders or alerts when key files or emails arrive
Tools to use:
Relay.app: epic UI, dead simple to use, great for structured workflows
n8n: flexible, can do nearly anything, requires 10% level of coding knowledge
Claude:
Make: solid drag and drop builder, better UX than Zapier
Gmail filters, Google Scripts: don’t overlook the basics
Things to Automate First:
Folder/file creation
Inbox clean-up and tagging
Document parsing (e.g. pull text from invoices, contracts)
Notifications + reminders
Meeting notes → summaries → tasks
You’re focused here on wiring things together rather than being a developer
3. Vibe Coding
Vibe coding = building scrappy, useful tools without waiting for a developer.
The name’s a bit silly, but it fits. It’s coding by feel. You describe what you want, AI gives you a starting point, and you fiddle until it works.
You don't need to know Python. But if you can:
explain a problem well
follow logic
edit what AI gives you then you’re 90% of the way there.
Things I’ve Vibe-Coded (or helped others create):
tracking expenses on holiday by auto logging purchases from bank emails into a travel budget sheet
building a tool to track investment performance from bank exports
making a doc parser that pulls key fields from documents into a table
building a persona builder for work projects
creating an audio and video file converter for different formats
Stack for Vibe Coding:
lovable, bolt: quick way to spin up simple apps without heavy lifting using natural language
google’s ai studio builder: easy to prototype ai features and connect them to tools
cursor: above my level, but I’ve hacked together an agent or two before throwing in the towel with it
kiro: ai-native ide that turns prompts into structured code, specs, and tests so vibe coding ends up clean and production ready. Above my paygrade
chatgpt, claude, or chatprd: great for helping with prds, breaking down prompts, and making vibe coding in the above builders more efficient
Vibe coding is about removing blockers for microservices or microsaas. You can build dashboards, calculators, tools, or internal scripts without needing a dev team. And when something breaks, you can read the error message and figure out how to fix it. That’s the real flex.
These Skills Stack Together
These three skills aren’t separate. They compound.
Fluency helps you think. Automation helps you scale. Vibe Coding helps you ship.
The people who build these are the ones who reduce their reliance on everyone else. You don’t have to ask the data team. You don’t need to wait for dev. You don’t need to brute-force through work manually. You can just do the thing.
Why It Matters in NZ
Kiwis are already generalists. Most of us grew up figuring things out ourselves. These tools are more tools for the toolkit, helping keep pace with global tools.
You can solve problems faster. You can ship ideas by yourself. You can run a business, side hustle, or internal ops stack without asking for help.
If you're feeling behind, just start with one:
Write a better prompt and improve the output
Pick a workflow to automate
Use AI to build a scrappy tool
Written by Mike ✌
Passionate about all things AI, emerging tech and start-ups, Mike is the Founder of The AI Corner.
Subscribe to The AI Corner
