We Sold a Generation a Dream and Dismantled the Ladder While They Climbed

I'm sitting in a pub in the City. Three separate interviews happening around me. Different tables, different companies, same conversation — layoffs, cut too deep, now looking for people with AI experience to manage their outsourcing in India. Three tables. Same story. Same morning.

That's not a coincidence. That's an autopsy.

The death of skillcraft doesn't look like empty factories or closed mines. Those were photographable. This is invisible — a skill never trained, a problem never owned, an engineer who became a project manager because that's what the market rewarded. You can't photograph an absence.

Post-2008 the calculus was simple — engineers are expensive, offshore the execution, keep the management. Rational at the firm level. Catastrophic at the civilisational one. Because you didn't just move the work. You moved the learning. The junior who grinds through a hard problem at 11pm and comes out the other side a senior — that stopped happening here.

What grew in the vacuum was middle management. An entire professional class whose incentive is not to finish the work but to extend it. Most London tech firms spend their time migrating, modernising, embarking on platform transformations that need redoing by the time they complete. The same stack, every three years. Not because the technology demands it — because when the migration ends, their purpose ends. So the migration doesn't end.

The product — the actual thing users need — became secondary to keeping the machinery of management justified and employed.

Senior salaries ballooned not because the value produced grew but because the complexity of managing the abstraction layers grew. Companies paid a premium for people who could navigate the bureaucracy that those same companies had built. The market turned. Those salaries look very different when growth isn't guaranteed. Companies cut again and discovered they cut muscle last time and are now cutting bone.

The third justification is being written right now. Why hire juniors when AI can fill the gap? Same logic as outsourcing. Faster, cheaper, better interface. Another nail in a coffin that was already being built — because this isn't new. In 2010, entry level positions required two years of experience. Entry level. The catch-22 was written into the job spec from day one — you cannot get experience without the job, you cannot get the job without the experience. The 2008 crash didn't just trigger outsourcing. It blocked the bottom of the pipeline the same day it deleted the top.

Some of them did everything right anyway. Bootcamps, degrees, side projects, GitHub profiles maintained like gardens. And then they entered a market that had offshored the training ground, built a management class with no interest in being replaced, and is now automating the justification for not hiring them at all.

A broken pipeline compounds. No juniors means no future seniors means no next wave of companies. No next wave means no success stories, no exits, no new angel investors, no founder who comes back and funds the wave behind them. Success stories are infrastructure. We don't treat them that way.

And the kid in a bedroom somewhere who looks at this industry and concludes the ladder was never meant for them — they're not wrong. Not yet. But they could be.

The big companies will move slowly. The bureaucracy that caused this will resist the solution to it. Don't wait for them.

If you're building something new — take a junior along for the ride. Give them hard problems and real ownership. Not as charity but because the company you build will be better for it. Law firms have known this for generations — a newly qualified solicitor costs more than they generate, absorbed consciously as a cost of survival. A firm with no pipeline is a firm one generation from extinction. The knowledge walks out when the partners retire and there is nothing behind it. Tech needs to make the same calculation.

AI makes juniors more capable than ever — use that. But AI will not produce the next generation of software architects. The instinct that comes from owning a system that fails at 3am, the judgment accumulated over a decade of hard decisions — that transmission is human and it always will be. It requires seniors who mentor, companies that fund the pipeline, and a community that stops pretending the next generation will emerge fully formed from a bootcamp and a Copilot subscription.

We built the tools. Now we have to build the people. That's not AI's job. It's ours.