The World Has Changed

Originally published by Dave Chapman on LinkedIn.

Last Tuesday my AI agent shipped a fully functioning travel-safety platform. Seventeen minutes of my time. Three hours of its time. A custom domain, Stripe integration, live on the internet. I didn't write a single line of code.

That was the moment I stopped thinking of AI as a tool and started thinking of it as a colleague - one that doesn't sleep, doesn't complain, and costs less in a month than a developer does in a day.

Go to sleep for six hours and wake up to find an endless amount of newly released developments and enhancements to catch up on. For the first time ever - as a lifelong technologist - I feel somewhat anxious.

For three months I've been deep inside OpenClaw, the open-source AI agent. Most of my tech friends have too. It's exciting, maddening, and occasionally magical.

Get it right and you feel like you're conducting an orchestra of infinite tools. Sometimes the magic is short-lived and something breaks, and it takes time to fix that issue and continue. It's time-consuming and addictive to the point of being genuinely unhealthy.

But one thing is certain. The world has changed.

Whether you are technically savvy enough to be using OpenClaw or have already jumped onto the consumer-ready competitor Claude Cowork, chances are you have come to the same conclusion. The ability to spin up a team of resources and complete tasks in minutes that have traditionally taken teams weeks or months to complete is profound. And the impact is going to be more far reaching than we can imagine.

Let me make this concrete.

TripTrust - my AI agent's first real creation - is a web service that provides in-depth personalised holiday planning and travel safety reports. In under three hours it was live. My own individual interaction was 17 mins, the rest was my AI agent. It deployed itself to the net using a custom domain. You can try it out here - TripTrust.co.

Using the same approach, in a few hours, I built ProtoWeave.ai - a business-idea validation tool that runs eight frameworks, generates financial plans, and even produces a pitch deck (message me if you’d like a promo code to try it out for free, rather than my suggested $29 USD fee).

Then there's the investment analysis platform: upload a data room (or just ask my bot to scrape it from those annoying, “secure” docsend websites) and in five minutes you get an analysis more thorough than a team of dedicated research analysts could produce in weeks. Here is an analysis of an AI fund I am considering investing in.

My agent has written over 25,000 lines of code in its last ten days, averaging over 2,000 lines of code per day. For context, most professional developers (in the pre-AI era) consider a day productive if they ship 50-100 lines of code.

These platforms are not perfect, not bug-free, and there is a real cost in terms of the LLM tokens they consume every time they are used. They also run on my development machine so the availability won't be production grade. I could make them production ready - but what's the point in polishing any of it, when honestly - all of this shit is going to zero.

The world is about to change faster and more radically than at any point in human history.

I know how ridiculous that sounds, and I do not say that lightly. Given where AI is today - and how fast it's advancing - I don't see how this is deniable.

I am fortunate to no longer operate a traditional technology business, but if I were to start one, I would be scrutinising which roles actually require a human to fill them. If I can spin up a team of agents that are skilled (most aspects better), don't sleep, don't complain and cost a fraction of a human - well other than social interaction benefits, why would I employ humans in roles no longer requiring them?

And if I were to employ humans in the roles that arguably don't require them, to help me feel better in an attempt to add value to society, a competitor wouldn't. Their subsequent costs of their services would be much lower. All of a sudden, the choice of hiring humans is no longer an option - it's business survival.

Here's where it gets genuinely strange. We're already seeing the emergence of agent-to-agent services - AI agents autonomously commissioning other AI agents for specialised tasks. My agent doesn't just write code; it delegates. It hires. Not humans, but other agents with specific capabilities.

The entire service chain from brief to delivery can now happen without a single human in the loop. This isn't a future scenario. This is happening on my computer right now.

Before I go further I want to be clear: I am not celebrating any of this. The people whose roles are at risk are not abstractions - they are colleagues, friends, and family members who built careers in good faith. The displacement that is coming will not feel like progress to those living through it.

I am sharing this not to alarm, but because I believe the worst thing we can do is pretend it isn't happening.

Based on my experience with OpenClaw/Claude Cowork since January, I've been thinking constantly about the impact this is going to have on white-collar workers in the coming months and years. We've started seeing the aggressive layoffs already in the tech and finance sectors (e.g. Jack Dorsey’s Block cuts 40% of staff, Atlassian cuts 1,600 jobs, HSBC mulls deep job cuts), but this is just the tip of the iceberg. I predict that most other sectors will follow suit very soon.

As much as technology excites me, I'm not rooting for people to lose their jobs - especially when finding another may be harder than at any other time in their careers.

I've been asked how many white-collar jobs I think will be impacted. Here's how I frame it:

• My conservative estimate: 30-40% of white-collar roles displaced or significantly restructured within 12 months - that's the scenario with regulatory drag and organisational inertia.
• My aggressive estimate: 50-60% within 18 months - that's competitive pressure forcing adoption at the pace the technology enables.

I think we land somewhere in between. And I think it happens faster than any of us are comfortable with.

The roles most immediately at risk are those built around information processing, content creation, basic analysis, and routine decision-making: paralegals, junior analysts, copywriters, customer support, junior developers, compliance officers, and middle management layers whose primary function is coordination.

These are not low-skill roles - many require years of training. That's precisely what makes this different from previous automation waves.

Everything above applies to white-collar knowledge work. But the timeline for blue-collar displacement is shorter than most people think.

I've ordered a Neo robot from 1X Technologies - not because I need a butler, but because I want to understand firsthand what's coming. Within 5-10 years, I expect humanoid robots to be handling warehouse logistics, basic construction tasks, and service-industry roles at scale.

The "learn to code" advice we gave to displaced factory workers is about to expire for everyone.

The Macro Picture

The macro consequences are only beginning to emerge.

Universal Basic Income shifts from fringe policy to economic necessity when 30-50% of the workforce is structurally displaced. Commercial real estate faces an existential reckoning as the offices required to house those workers become redundant.

And here's the counterintuitive one: nations with declining populations - long treated as demographic disasters - may emerge as surprising beneficiaries. Fewer people to displace, lower social strain, higher GDP per capita as AI and robotics amplify output. Japan and South Korea, with their deep tech infrastructure and capital reserves, look genuinely well-positioned. The caveat is real, though: aging populations carry massive pension and healthcare obligations, and the productivity gains of AI only matter if they're distributed broadly rather than captured entirely by capital owners. It's a better hand than most people think - but only if it's played right.

So What Do You Do With This?

I'm not going to tell you to "upskill" or "learn prompt engineering" - that advice has a shelf life of a few weeks at best, and frankly it misses the point. The point is that the relationship between work and value is being rewritten in real time, and most are still only using AI as a chat prompt.

What I'd actually suggest: get your hands dirty. Don't read about AI - use it. Build something pointless with it this weekend. Break it. Understand what it can and can't do, not from a headline but from your own experience. The people who will navigate this best are not the ones with the most credentials - they're the ones who stopped pretending it wasn't happening and started experimenting while there was still time.

And if you're in a position to influence how your organisation adopts this technology - whether you're a founder, a manager, or just someone with the ear of people who make decisions - push for honesty. Not optimism, not panic. Honesty. Because the conversation we should be having isn't "will AI take my job?" It's "what does my job look like in a world where AI does the parts I thought were the job?" That's a harder question. But it's the right one.