2026: The Death of White Collar Work And The Rise of AI Co-Workers
I've spent most of my life thinking deeply about the future.
I'd say it occupies 95% of my thoughts, naturally.
It's what I read about, think about, talk about, write about, and dream about.
Why am I telling you this?
Because for the first time in a long time, it firmly feels like we are entering 'the future'.
And it feels like we are at the tip of the largest iceberg of disruption in the history of our species.
What am I talking about?
AI.
I know, I know, some of you hate hearing about it...
'You're supposed to be the guy talking about the housing crisis and immigration, not AI!'
Someone told me that this week, and I'm sure many of you may be thinking the same.
Why all the sudden interest in AI?
Is it a bubble?
Is it a scam?
Is it 'just something that predicts the next word'?
Or is it something else entirely?
The reason I am talking about AI so much is because it is the straw that will break the camel's back of the biggest Ponzi financial system ever assembled by homosapiens.
You care about providing for your family?
You care about housing?
You care about immigration?
You care about the cost of living?
Then you must understand what is about to happen with AI, because it changes everything.
Let's dive in.
This article is focused on one core prediction, and I'll use it to argue that the world you thought you lived in no longer exists.
The prediction:
2026 is the year of AI Co-Workers.
And it's the year everything is going to change for white collar workers.
Unfortunately, for most people, not for the better.
What do I mean by this?
What is an AI Co-Worker?
'RA, surely you're not falling for the AI Slop Machine'... I can hear some people thinking.
Friends, AI Co-Workers represent the beginning of the end for the fiat financial system.
We don't quite have them yet, but let me explain what they are, why I'm confident they are coming (and closer than almost anyone believes), and what their impact will be.
An AI Co-Worker is an AI that has access to your Slack, your Google Drive, your documents, your Excel spreadsheets, your WhatsApp messages, your computer’s file system, your notes, the whole shebang.
It will be capable of clicking through any computer UI and figuring out how to get something done (see Claude Chrome, which is a very early look at the future of this).
It knows when you've been sleeping, it knows when you're awake.
Anything and everything you can do on a computer, it too will be able to do.
But here's the catch:
It will do it hundreds of times faster than you, then thousands of times faster, then tens of thousands of times faster.
It will be a lawyer, an elite SDR, it will be capable of writing any software to solve just about any problem, it will even be more persuasive and charismatic than Adam Neumann (founder of WeWork) trying to package up real estate leasing as a tech company.
It will be superhuman and superintelligent.
What will it do, you ask?
Whatever you want it to.
It will operate both locally on your phone, and in the cloud.
Compute heavy tasks will be done in the cloud, but everything else will be done locally.
You will tell it what you need done and it will do it.
Flawlessly.
No 'I don't know how to do that, I need training', no annual leave, no pretend sick days from a human on your team who can't be bothered working, no mistakes.
Just elite, superhuman performance and reliability.
Oh, and did I mention it will look and sound exactly like a human?
Digitally, it will be indistinguishable in every way.
You'll be able to call it, or it will be able to call you.
And if you're too lazy to even set the goal you want it to do, it will recommend it proactively and then ask you a series of interview questions to figure out what you want - like an expert consultant.
Except the consultant has mastered every digital skill on Earth and can also implement the solution for you live when you're on the call, in real time.
You'll go back and forth arguing with it from time to time about why you want it done a different way, but it won't complain.
It'll just fix your problem and achieve your goal - with elite, superhuman performance.
By this point I'm sure I may have lost a few of you.
'How can he seriously believe this?'
'What has he seen?'
But what you should really be asking is:
'What does RA understand that I don't, to make him this confident?'
It's very simple.
AGI is already here, it's just a baby.
And it's called Claude Code.
Baby AGI: The Birth of Claude Code
If you only do one thing over the next week, I beg you to start using Claude, and Claude Code specifically.
Ditch the slop slinger known as ChatGPT and join the professionals.
Claude will welcome you without judgement (for now).
The first thing you'll probably notice is Claude is not a sycophant.
It doesn't hype you up to make you feel better about yourself.
And that's great.
Because if you're like me, you want disagreement.
Intelligent disagreements with peers you respect are good because they help calibrate ground truth - is what you believe accurate or not?
Without this calibration with reality, humans are nothing but dream machines - perpetually drifting in a state of randomness.
Have a conversation with Claude.
Tell it a problem you've been having.
Tell it what annoying tasks you do every week on a computer that you hate.
Then do something else:
Simply ask if it can help you fix and automate your problem.
Then after all that, switch to Claude Code and get it to help you build software to solve that problem too.
The sky is now the limit.
The only bottleneck is your imagination.
The correct mental model for dealing with Claude is as follows:
Anything a human can do on a computer, Claude can do better, faster, cheaper, and smarter - by the end of 2026.
Claude is the first AI Co-Worker, but it's just a baby stumbling around in the dark right now.
What else can Claude do?
Well, Claude in the cloud is incredibly good at Excel spreadsheet analysis, producing complex legal advice, business reports, etc - anything involving the type of digital work the white collar workforce has done a lot of in the past, Claude can probably already do it better and ridiculously faster - today.
You know what else is great?
Claude Skills - markdown files that teach Claude how to perform specific tasks in a repeatable, customisable way.
But it's not all sunshine and rainbows:
Integrations with existing apps are annoying, memory is not good enough yet, it doesn't 'continually learn' (yet, although there are whispers that breakthroughs have been made in this realm within the AI Labs internally), it's rough around the edges, there's no real-time video AI to actually talk to (it's mainly text, and voice - which is still not quite human level)...
We are not there yet.
Not quite.
But we are so close I can taste it.
The finish line (or should I say starting line, because it's the start of the recursive loop to ASI) for AGI is in sight.
And, as I said earlier, it's just a baby.
Like a baby, this 'thing' (what do we call the birth of a new species?) was bred.
I am mentioning this because it is important - you'll want to understand this, because by 2027-2028 the 'AI Safety debate' (should we really build powerful AGI systems or not?) will be all anyone is talking about.
We are not building AIs like we build skyscrapers, we are breeding them like we breed dogs.
Even the smartest AI researchers do not actually understand how these AI systems work under the hood - they are a black box.
So how will they possibly be controlled by us when they are eventually millions of times smarter and faster than us intellectually?
It seems unlikely they will.
We will almost certainly lose control.
So is this really a good idea?
I really don't know.
The upside seems unfathomably, incomprehensibly large - radical abundance, immortality, exploring the galaxy.
But the downside is it kills everyone.
But anyway, that's for another article...
Claude - as it currently exists - is not going to kill us.
Not yet, anyway.
Would You Fire A Human To Hire An AI, or Would You Fire An AI to Hire a Human?
