A year ago, I posted a video titled “The Death of the Software Engineers.”
And I remember the pushback clearly: “Nah… this isn’t going to happen.”
I said it would — and honestly… the last 12 months have been the receipt.
In a single year, we’ve gone from “AI helps you write functions” to agentic coding that can plan, execute, debug, and work through larger codebases with way less hand-holding.
Just look at the pace:
•OpenAI shipping GPT-5.3-Codex as a faster, more capable agentic coding model, and expanding the Codex experience with an app built around multi-agent, long-running work.
•Anthropic upgrading Claude Opus 4.6 with stronger coding debugging, longer sustained agentic tasks, and even a 1M token context window (beta).
•GitHub rolling out GPT-5.3-Codex inside Copilot — meaning these capabilities are showing up right inside dev workflows, not as a “cool demo.”
So no, software engineers aren’t “dead.”
But the old definition of the job? That’s what’s dying.
The advantage is shifting hard toward:
•problem framing (turning chaos into specs)
•architecture tradeoffs
•tests, reliability, security
•shipping outcomes, not just code
Here’s the original video (watch it with 2026 eyes):
youtu.be/R36Ck3z6Ma8?si=9rZE…
What’s one task you used to do manually a year ago that you now delegate to AI?
And this is exactly why I built Prossess MVP Studio: to help founders and teams ship fast with AI-accelerated engineering while everyone else was “at the beach.”