
Is Automation Dead? I'm Starting to Wonder.
I built a business teaching automation. Now I'm using it less than ever. A refund request made me stop and ask why - and whether the tools I used to swear by still have a future.
Read Story→
Let's start with the data, because it landed just last week and it's hard to ignore.
The Stanford Institute for Human-Centered AI released its 2026 AI Index on April 13th (a 400-page annual report on the state of artificial intelligence). Most of the headlines focused on AI model performance and the US-China race. But buried in the employment section was something that should have made the front page.
Employment among software developers aged 22-25 has fallen nearly 20% since 2024. Meanwhile their older colleagues' headcount has grown. The decline is targeted, not general. It is happening to junior developers specifically, in roles with the highest AI exposure, while senior and mid-level positions hold steady or increase.
featured
Join 500+ founders who are vibe-coding their customer support with Relavo.
Try Relavo FreeThis isn't an isolated finding. A recent McKinsey report on technology workforce design published this month states plainly that "the business case for hiring large numbers of junior developers is weakening as agentic AI absorbs much of the work that once justified those roles." Salesforce CEO Marc Benioff confirmed the company stopped hiring engineers last year. The picture is consistent across credible sources. The junior developer role is under serious pressure.
The reason isn't complicated. AI tools now handle the tasks that used to be assigned to junior developers - writing boilerplate code, debugging, documentation, basic functions. Work that once kept a junior busy for a day now takes a senior team member ten minutes with Claude or Copilot.
The economic logic for hiring a £60k graduate to do what an AI does instantly has collapsed. Companies that previously hired five juniors now hire two mid-level engineers with AI tools and get comparable output.
Here's what concerns me more than the current numbers.
If juniors can't get jobs, they never become mid-level developers. If mid-levels are being compressed, there's nobody growing into senior roles. The career ladder depends on people climbing it. Right now the bottom rung is being removed.
Ten years from now, who reviews the AI-generated code if nobody got hired as a junior in 2025 or 2026?
Junior developers are struggling right now. The data is clear. But I'm not convinced this is permanent - and the reason why comes back to something I've been writing about on The Vibed.
Vibe coding has made it possible for non-technical founders to ship applications fast. That's genuinely exciting. But most of those founders don't fully understand what's happening in the code that AI is generating. We're relying on it being correct, being secure, being maintainable.
And increasingly, it isn't.
Research shows that up to 30% of AI-generated code contains security vulnerabilities. Code that gets rewritten or deleted within two weeks because it doesn't work properly has doubled. We are heading towards a marketplace flooded with vibe-coded applications that look polished on the surface and are held together with digital duct tape underneath.
When that starts to matter - when the security breaches happen, when the scaling problems hit, when the technical debt becomes unpayable - someone is going to need to fix it. That someone is going to need to understand code.
Right now, the entry-level market is brutal and I won't pretend otherwise. The Stanford data is fresh and the trend is real.
But CNN Business reported just two weeks ago that job openings for developers are actually growing - because companies believe they can produce more software now that nearly anyone can build with AI, increasing demand for engineers who can shape and oversee those products.
The role is possibly evolving more than it's dying. The juniors who will thrive aren't the ones who can write boilerplate faster than AI - nobody wins that race. They're the ones who understand what the AI is actually doing, who can read the output critically, who know when to trust it and when to push back.
The vibe coding wave is creating a lot of surface. Someone's going to need to worry about what's underneath it.
Thinking about the risks of building with AI? We covered the security side of vibe coding in an earlier piece - worth a read before you ship.
More from The Vibe Check

I built a business teaching automation. Now I'm using it less than ever. A refund request made me stop and ask why - and whether the tools I used to swear by still have a future.
Read Story→
Vibe coding is one of the biggest opportunities available to entrepreneurs right now. But there's a side to this that most people aren't talking about.
Read Story→