A while back
In the early days of Facebook – the very early days – I walked into one of New Zealand’s major gyms and saw them celebrating on a whiteboard how many followers they had on their Facebook page.
I told them I didn’t understand why they were bothering. “This social media thing,” I said, “I’m not sure it’s going to take off.”
They still remind me about that conversation whenever I see them.
So look, take what I’m about to say with that in mind. I’ve been spectacularly wrong before, and I could be wrong again. But here’s a framework I’ve been mulling over that’s helping me make sense of what AI might actually disrupt – and when.
It’s About the Interface, Not the Intelligence
Most predictions about AI replacing jobs focus on whether work is “creative” or “routine,” “complex” or “simple.” Honestly, I don’t think those distinctions are very useful anymore. AI is writing novels and struggling to fold laundry.
What I’m finding more helpful is thinking about interfaces. Specifically: how does the work connect to other systems and people?
My guess is that AI’s ability to replace human work isn’t really gated by how smart it is. It’s gated by how mature the interface is between AI and whatever it needs to interact with.
Three Layers (As I See Them)
File to file – This is where AI is already operating at full strength. Data in, data out. No ambiguity, no emotional nuance.
Coding fell first because it’s essentially file manipulation.
The AI reads files, writes files, and never needs to convince anyone of anything. I’ve seen this firsthand with my own agency – we’ve built automated workflows that handle tasks which used to take hours, though we still have humans checking the output to make sure quality is right. It’s not replacing the team so much as leveraging what they can do.
If your work lives primarily in this layer, AI isn’t coming. It’s here.
Asynchronous communication – Email, chat, messaging. These interfaces are maturing fast.
Here’s the thing though – I don’t think adoption will be driven by people suddenly loving AI support. It’ll be driven by frustration with the alternative. Would you rather wait 90 minutes on hold for a human, or get an AI response in 30 seconds that sorts out 80% of your problem?
I know which one I’d pick.
Face-to-face interaction – Real-time, in-person, full bandwidth human communication. Reading body language, building trust through physical presence.
This is the frontier, and not because AI couldn’t theoretically handle it – the interface technology just isn’t there yet. Roles that are heavy on in-person interaction have more runway.
But I’d be careful about confusing “more runway” with “safe.”
What Happens to Jobs That Span Multiple Layers?
This is something I’m still thinking through, honestly.
Most jobs aren’t purely file-based or purely face-to-face. A project manager works in spreadsheets but also runs meetings. An accountant processes data but also advises clients over coffee.
My sense is that the file-based portions get automated first, which concentrates the job into its human-interface components. That sounds like protection until you realise it might also mean fewer hours of billable work, or the same work spread across fewer people.
I’m not certain about this, but it’s worth considering.
The Forcing Function
Here’s where I do feel more confident: none of this happens because customers demand it or businesses are eager to change.
It happens because someone in your industry decides to build the 80/20 model – 80% AI, 20% human, at a fraction of the current price.
Once that competitor exists, customer hesitation becomes a luxury no one can afford. The company that said “our clients prefer the human touch” either adapts or watches the market shift.
This is probably why disruption feels gradual until it’s sudden. Years of debate, plenty of “people will never accept AI for this” takes, and then one business proves the model works.
Some Questions Worth Sitting With
I’m not going to tell you what to do – I got Facebook wrong, remember? But here are the questions I’ve been asking myself:
What interface does my work primarily flow through? Files and data? Email and chat? Face-to-face?
For the parts that are file-based or async communication, am I using AI to do more with less? Because if I’m not, someone else probably is.
And maybe the most uncomfortable one: who in my industry is going to make the 80/20 move first? If I don’t know, is there an argument for it being me?
I don’t have all the answers. But I reckon the framework is worth thinking about.
One Thing That Won’t Change
Whatever happens with AI – whether it’s handling your emails, your data, or eventually your customer calls -one thing stays constant: strong messaging wins.
It doesn’t matter if your marketing is delivered by a human or through digital channels or eventually by AI. Weak messaging loses to strong messaging every time. The businesses that get clear on what they’re actually saying to their customers will outperform those that don’t, regardless of what technology is delivering that message.
If you’re wondering whether your marketing messaging is as sharp as it could be, we offer a messaging audit that takes a hard look at what you’re currently putting out there. Sometimes a fresh set of eyes spots things you’ve been too close to see.
Get in touch if you’d like to chat about it, call or use the form below.
