Why Autonomous AI Needs Autonomous People First

You’ve been given a mandate at work - excel with autonomous AI.

The only problem is that no matter how much you learn, automate or build… you feel like you’re dancing in a circus.

A win that should have bought you weeks, gets swallowed up by one approval cycle. 

An automated process that once took 3 hours, has added 5 hours of new meetings to your schedule in automation reviews. Your end-to-end agents aren’t allowed to run together because of security vulnerabilities. 

Or token use. Or Sandra in compliance.

• Microsoft says meeting times are WAY up, and 16% higher in the evenings.
• 79% of workers say they can’t focus at work for more than an hour.
• Research from UC Irvine says we have 12 mins on a task before interruption.

There’s something at work keeping you stuck in the AI kiddie pool.

Your company leaders will say it’s because you need to upskill. Or find new tools, or test new workflows. They’ll say it’s because you aren’t working hard enough to find solutions. After all, CTO Greg has 457 agents doing his job for him!

But those are all convenient lies.

The truth is that you’re not allowed.

The office with its permissions culture and red-tape bureaucracy is where advanced AI innovation goes to die for the average employee. With the jarring rise of the infinite workday, work is spilling over into mornings, commutes, evenings and weekends.

Harvard has written extensively about how AI can intensify work. That’s a vanilla ice-cream way of saying AI often creates harder, bigger workloads. And more of them.

I was reading studies and research on the rise of AI superagency last week, reviewing who is benefitting from AI the most – and it struck me.

We want autonomous AI. But we don’t want autonomous people.  

Everyone in tech is chasing perfect AI autonomy in workplaces where people aren’t even autonomous yet.

It’s time we faced the cold hard facts folks – the office inhibits AI innovation.

And it’s because AI can’t be autonomous in a workplace that won’t allow people to be.

This autonomy monkey on all of our backs needs to be called out into the open – or it will construct a career ceiling that will keep you from growth, progress and getting ahead in your industry.

In this article, I’m revealing an uncomfortable truth - the blocker isn’t your skill, it’s your environment. I’ll show you how autonomy drives advanced AI use, who is already harnessing that power, and how to protect your career in a job market that rewards self-directed operators.

The Office... Not an Autonomous Place

Ah, the old 9-5.

I mean, 7-7 if you have a long commute. Sorry, what I really mean is 5-9, if you’re working in a highly competitive AI-focused tech industry.

Not long ago, hundreds of thousands (perhaps millions) of remote workers were forced back to the office because of ‘collaboration, creativity and productivity’ concerns.

It definitely had nothing to do with empty office space, real estate losses or leaders feeling an absence of control over their workforces.

RTO mandates ran rampant globally, with most settling on a hybrid model.

Our friend Dr Gleb Tsipursky - wrote about the narrative that surfaced – claiming that without the office, innovation dies.

So remote workers reluctantly left the happiest moment in their career history, to commute back to the office as instructed.

Then the AI boom happened.  

A period of time that has accelerated and accentuated growth or decline. Like remote work, it’s revealed who the real top performers are. And it’s shown – quite plainly – that the previous arguments against remote work, were all forms of corporate gaslighting.

Here’s what we know:

• 87% of leaders use AI on the job vs 27% of employees (2025)
• Leaders work remotely more often than employees, even senior ones.
• Leaders have much higher levels of autonomy than the average employee.
• US has the most AI funding, development and enterprise AI integration.
• Yet lags in terms of workforce adoption

The US workforce has much lower levels of autonomy in the office. It’s almost as though the individuals who have more discretion and freedom over tools, tokens and testing – are the ones streaking ahead with AI technology.

The pattern here is clear – the US is building incredible AI in an environment where autonomy is rationed. This happens in many other places in the world too.

Office culture is asking for permission to take a day off. To go to the doctor. To have a holiday or respond to a family emergency.

And even then, employees are at the mercy of their managers, who are often extremely incentivized to control risk, not unlock autonomy.

The bottleneck isn’t you or the AI. It’s your environment. The very culture they hurried you back to for ‘the big RTO or get fired’ push of 2024.

You’re expected to build autonomous AI, in an environment where autonomy doesn’t exist. Not for you, or any of your colleagues.

It’s always going to be an uphill battle of trust.

Permission loops, endless approvals and ongoing fear around accountability will keep you from your full AI potential.

Here’s something you might not know.

We know remote leaderships works. We also know that remote workers are the ones doing insanely cool things with AI.

Because remote workers are autonomous people. And office workers aren’t.

The Autonomy Monkey and Remote Work

I wrote an article about AI and task removal a while ago. In it I outline how AI progressively removes tasks off your plate, so that you can focus on bigger ones.

But what happens when you can’t build freely with AI at work?

I genuinely believe this big new task office tech workers have been given – to ‘create autonomous AI processes’ is a little bananas.

It reminded me of the story of the monkeys in The One Minute Manager Meets the Monkeys, by Ken Blanchard, William Oncken Jr and Hal Burrows.

The book explains how to prioritize tasks and keep your time focused on the right things. Inside, the authors tell us how monkeys are ‘your next steps’ at work and how they can intensify and end up overwhelming you.

The lesson of the book is ‘don’t take on the problem if the problem isn’t yours.’

At the office, many of us take on problems that don’t belong to us – accumulating more monkeys. Work doesn’t flow to the person closest to the problem, it flows to the person with the title, approval power, risk tolerance and the final say.

Basic human autonomy is suppressed at work, so it’s not surprising that advanced AI autonomy is a glorious fairytale.

The Autonomy Monkey is real – it’s the work you could take off your back, the decisions you could make and the workflows you could run…but you don’t.

Because you can’t. You’re not allowed too.

In Microsoft’s grand journey to the frontier firm, they speak about phase 1, when people use assistants, and then phase 2 when people use human-agent teams. They barely consider the reality that phase 3, human-led, agent operated AI is going to be a massive problem in most organizations.

[source]

How are we supposed to trust people to build AI, when we don’t trust them in the first place?

Tragically – that means finding optimal balance with the right human-agent ratio remains elusive for most tech workers – trapping them helplessly in the imbalance – when they have too many single agent systems to operate.

When you train people to operate dependent on YOUR permissions, they will train AI to be dependent on permissions too. You swap autonomy for dependence and end up creating a mess of scale.

It’s ingrained in office culture.

The Remote Work Autonomy Upgrade

You know who has a PHD in autonomy? Most remote workers.

We had to learn how to become monkey managers to sustain our careers. There is no time for performative busyness, when metrics are all that matters. 

When results are the reason you get to stick around for another quarter.

We don’t get interrupted often. We’re masters at deep work. And we have god-mode levels of self-responsibility. Moving work forward is literally what we do.

Could it be that remote work autonomy facilitates AI advancement?

Over time, working remotely creates a different skill set - ownership, prioritization, delegation, and the ability to act without waiting for someone else to bless our decisions.

And the evidence suggests it’s why remote workers are crushing advanced AI.

If you’re in tech at the office every day, working harder than ever with AI – and not getting very far…it’s not you. You’re not failing. You’re trying to create unnatural autonomy where it doesn’t exist. It’s a catch 22.

If you need approval to move your work forward, the best agent in the world is just going to become a colossal monkey on your back, permanently waving a permission slip at you. Autonomous AI can’t thrive in a permissions-loaded environment.

But it can survive (and scale) in a remote environment, where autonomy IS the culture.

Here’s how.

#1: AI Doesn’t Create Autonomy (It Reveals It)

I once called AI a mirror, an accelerator.

In a workplace environment where you’re trusted to act, AI is serious leverage. But the opposite is also true – in most challenging office environments, AI can be a nightmare.

Or more accurately, a scope and scale circus.

Microsoft has told us where agent-operators are heading – and it’s clear that this level of AI only works when people are trusted implicitly and allowed to take the lead.

[source]

You need to know that AI doesn’t magically turn you into an autonomous operator. It reveals whether you’re allowed to be one.

• When AI gives you ‘the next step’ monkey – can you take it?
• Can you ship without 100 meetings?
• Can you change, update or improve your workflow without approvals?
• Can you connect tools, run processes and assess risk without someone panicking?

If the answer is no, your problem isn’t AI – it’s autonomy.

AI adoption and skill building in this kind of office environment is going to leave you with a stack of monkeys on your back. Tools can do the work, but if you can’t effectively operate the tools – you’re never getting to phase 3.

Recognize that this is a career ceiling and smash through it.

• Find a lane you fully own (a workflow, metric, system). Use AI to run the monkeys inside that lane to get ahead.
• If your current job won’t let you offload your monkeys using AI, leave.

Take stock of how much autonomy you have and whether it’s helping you succeed or holding you back in your current role. Then take decisive action now.

#2: Permission Culture Creates (Bad) Dependent AI

Now that you’ve figured out if you have real autonomy or not, let’s translate that.

According to a whitepaper from Herman Miller, autonomy is a fundamental human need. When you have it at work, you’re more productive, creative and innovative!

Innovation is a byproduct of being autonomous.

Office-based permission culture destroys autonomy, and as a result – it actively suppresses innovation at work.

• Autonomous AI work is experimental BY nature: Great agents and workflows require iteration, judgement calls and super-fast feedback loops.
• Autonomy makes those loops possible: Ever tried to iterate something without permission? It doesn’t work.  
• Autonomy is the pre-condition for AI-driven innovation!

Your team might think they’re creating autonomous AI, but permission culture distorts builds and you end up creating dependent AI (and burnout people). You’re forcing AI into assist mode instead of operate mode – kind of like what your boss does to you.

Evidence exists across 16,000 employees from 17 countries that organizations who grant people freedom over how and where they work perform better – autonomy-forward companies outperform others by 17%.

If you’re using AI badly or building AI flows that cause team burnout, you need to get out of dependency mode: kill the approval loops, give someone real ownership, and let the agents actually do the work.

#3: Remote Workers Win Because We Own the Next Step

I still laugh about the old ‘remote work makes productivity worse’ conspiracy theory. For high performers remote work freed them to be 10X more effective.

And now AI has done the same.

It’s not co-incidental that remote workers are the most autonomous tech workers that exist. Remote work lends itself to the kind of environment advanced AI use needs to reach its maximum potential.

A Stanford/SIEPR working paper says that AI has a much higher impact in the home than at the office.

This is because folks working from home have the autonomy to integrate AI into their daily flow - across professional and personal tasks.

With ZERO corporate oversight strangling the workflow.

So -

• Office workers have to apply AI like a rigid policy
• While remote workers can use AI like a system

The difference shows up in innovation, not just stuff being built and created.

Once you can offload the autonomy monkey work, you get something rare in return - extra cognition. And any spare cognition is where experimentation lives.

This is when you start testing mad ideas. You iterate at blinding speeds. You simplify workflows into elegant symphonies of action. You ship incremental improvements instead of just generating more and more draft versions.

Remote work was always the advantage – but now it’s the key to AI superperformance.

If you want to crush it with AI as a remote worker, the rule is:

Own the next step. Then hand the monkey work to autonomous AI.

• Ownership: “What outcome is mine?”
• Monkey-spotting: “What’s sitting on my back?”
• Delegation: “What can AI carry?”
• Leverage: “What do I build with the brainpower I just got back?”

That’s the autonomy upgrade. Remote workers aren’t winning because we have better tools and more time to use them. We’re winning because we’re allowed to move the work forward – and prepared to handle the consequences of risk.

Stop Asking Permission to Be Innovative With AI

A lot of people are feeling disillusioned with AI right now - I get it. Truth is, autonomous AI needs autonomous people first.

Folks who can take the next step without summoning a committee. People who can test, iterate, and ship without turning every experiment into a risk hearing.

If your job doesn’t trust humans with autonomy, it will never trust your agents either. So, your autonomous AI journey won’t end with innovation.

It’ll end with you carrying more monkeys, producing more AI output, and getting evaluated on results you were never allowed to deliver.

Stop trying to innovate harder inside a culture that punishes autonomy. That’s not grit, that’s self-sabotage.

Better prompts and bigger agents won’t save your career. Better conditions will.

A remote environment where autonomy is the default, where you own the next step, where AI can operate without begging Sandra in compliance for a hall pass.

It’s time to quit this circus.

And go where the monkeys get off your back, for real.

Find an autonomous remote job here.