If you’ve ever watched someone use a piece of business software, you’ve probably seen the same thing I have.
Click.
Next.
Save.
Back.
Search.
Open.
Edit.
Save.
None of those clicks are the work.
They’re simply the translation between what the person wants and what the computer can understand.
For decades we’ve accepted that translation as a fact of software.
I’m no longer convinced it is.
We designed interfaces to solve one problem.
Computers couldn’t understand us.
If I wanted to mark an invoice as paid, I couldn’t simply tell the computer.
Instead I had to navigate the application.
Find the customer.
Open the invoice.
Click “Receive Payment”.
Enter the amount.
Choose the bank account.
Press Save.
The interface wasn’t there because humans enjoy clicking buttons.
It was there because the computer couldn’t understand:
“John has paid invoice 123.”
The interface was translating my intent into something the computer could process.
Generative AI changes that.
For the first time, computers are beginning to understand intent directly.
Not perfectly.
Not always correctly.
But well enough that the translation layer is no longer the only option.
That raises an interesting question.
If the computer already understands what I want…
…what is the user interface for?
I don’t think user interfaces disappear.
Far from it.
I think their job changes.
Instead of translating human intent…
…they become windows into what the computer is doing.
Imagine asking:
“Publish version 2.1.”
I don’t want a wizard.
I don’t want six configuration screens.
I don’t want twenty checkboxes.
I want to watch the behaviour.
Generating release notes…
✓ Complete
Creating GitHub release…
✓ Complete
Running integration tests…
✓ Complete
Waiting for approval…
The interface hasn’t disappeared.
It has changed from input…
…to understanding.
The same thing applies almost everywhere.
Think about booking a holiday.
Today we navigate websites.
Search.
Filter.
Sort.
Compare.
Adjust dates.
Open another tab.
Repeat.
What we’re really trying to say is something much simpler.
“I’d like a relaxing week in the sun in September for around £1,500.”
The rest is translation.
Or think about restaurants.
Recently I found myself waiting to pay for breakfast before catching a flight.
The restaurant already knew everything.
My table.
My order.
The total.
The VAT.
The only thing missing was a way for me to express one simple intent.
“I’d like to pay my bill.”
Again, the interface wasn’t helping me think.
It was helping the restaurant understand.
There are still things interfaces do remarkably well.
Humans are visual creatures.
Charts reveal trends that paragraphs never could.
Maps show relationships.
Timelines explain history.
Dragging a crop rectangle is often easier than describing one.
Visual editing isn’t going away.
Neither is exploration.
If anything, those parts of software become even more important.
What disappears is the endless collection of forms whose only purpose was translating people into data.
Dropdowns.
Checkboxes.
Wizard screens.
Nested menus.
Search boxes.
They solved a problem that may no longer exist.
This is one of the ideas behind Vork.
Not replacing every interface with a chatbot.
That would simply be replacing one poor interface with another.
The real opportunity is to separate behaviour from presentation.
The behaviour doesn’t care whether it was triggered from Slack.
A web browser.
A mobile phone.
A kiosk.
A voice assistant.
Or something we haven’t invented yet.
The behaviour stays the same.
Only the window into it changes.
I don’t think we’re witnessing the end of user interfaces.
I think we’re witnessing the end of user interfaces whose primary job was translating intent.
For fifty years we’ve built software to help humans communicate with computers.
Perhaps the next fifty years will be about helping humans understand what computers have already understood.
That feels like a much more interesting future.