Coffee and Code: Is AI actually saving you time, or just putting lipstick on a pig?
In the last 12 months, I’ve built a global SaaS platform and pivoted an IT consultancy, essentially solo. When you are a team of one, you look for leverage wherever you can get it.
For decades, that leverage was coffee. In 2025, it was AI.
Interestingly, they share the exact same risk profile.
I have a strict routine involving my Breville Espresso Machine (she’s the real MVP of the office). One double-shot espresso focuses the mind. It’s rocket fuel.
But during the long nights of coding, one cup became two, then four. I wasn't buying energy; I was buying anxiety. I remember sitting at my desk, physically vibrating but mentally unable to execute a single task.
Artificial Intelligence follows the exact same curve. Used strategically, it’s a superpower. Used blindly, it’s just a faster way to create a mess.
The "Speed Trap" (And why Kiwis are at risk)
We are adopting this technology faster than almost anyone else. Recent data shows that organisational adoption of AI in New Zealand hit 87% in 2025 (a massive 32% jump from the previous year). Compare that to the USA, where only about 40% of the workforce is really using it.
We are rushing to the oracles (ChatGPT, Gemini, Perplexity) for answers.
And businesses are rushing to plug "AI Chatbots" and "Auto-Marketing" tools into their websites, hoping it will be the "Easy Button" for efficiency.
But here is the question you need to ask: Is the AI actually saving you time, or is it isolating your customers?
When the Model Lies to You
AI models are moving fast. Sometimes, too fast for their own good.
During my own development work building Koru Compass (an agent that reroutes guests based on live weather), I hit a wall. I queried a leading AI model for the coding protocols required for its own latest update.
It gave me outdated information. When I challenged it, it got confused, apologised, and gave me different wrong information.
I’m a developer. I knew how to fact-check it against the manual documentation.
But what happens when that "confident hallucination" happens to your customer?
The "Hallucination" Hangover
Imagine you install an AI chatbot on your business website to save admin time.
It tells a customer you are open on Waitangi Day (when you’re closed).
It promises a discount code that expired three months ago.
It recommends a product you haven't stocked since 2023.
That doesn't save you time. It forces you to spend hours apologising to angry customers. It doesn't build relationships; it isolates them.
Data Sovereignty: The Boring Secret to Success
Businesses are expecting AI to fix everything, but they are ignoring the foundation: Your Data.
AI is just a multiplier.
If you multiply Good Data × AI = Efficiency.
If you multiply Bad Data × AI = Automated Chaos.
When I built the guest agents for my SaaS platform, the hard part wasn't the "AI" part. The hard part was the Data Sovereignty. We had to ensure that the underlying records—locations, opening times, advertiser relationships—were 100% accurate before we let the AI touch them.
If your core data is messy—if your inventory is tracked on sticky notes or your customer list is in three different spreadsheets—plugging in an AI tool won’t help. It will just confidently lie to your customers at scale.
Friend or Foe?
AI is a friend, but it requires supervision. It is not a replacement for human expertise or clean records.
If you are thinking about deploying AI in your business, stop looking at the shiny tools for a second. Look at your spreadsheets, your CRM, and your process documents.
If those are a mess, the AI will be your foe.
If you want to get your data ready for the future, that’s where we start. Not with a robot, but with a plan.
Otherwise, to borrow a phrase that might trigger some election-year flashbacks... you’re just putting lipstick on a pig. 🐷