Over the past few months I completely rebuilt my approach to software development.

I used to write code myself. Now I strategically manage neural networks.

Before, I spent time on architecture, reading documentation, and understanding other people's code. Now I spend my time choosing the right model and discussing on Twitter which one is the real game changer.

I bought all the most expensive AI coding tool subscriptions. About $500 in total. Honestly, I feel sorry for people who haven't done this yet. I'm subscribed to all the founders of AI dev tools and read their posts religiously. If someone claims their product can replace a mid-level developer, I see it immediately. You have to stay ahead of the market.

I used to worry when I didn't understand some complex part of the system. Now I usually don't understand what the neural networks wrote for me — but that's fine. I read on Twitter that reading and understanding code is becoming obsolete. Modern engineers don't need to understand code; they orchestrate agents.

If the code compiles — great. If it doesn't — I just rephrase the prompt.

I no longer create tickets in Jira myself. I do it through Claude. Yes, it takes longer, but at least I'm using the latest tools. First I ask it to formulate the task, then I refine the context, then I ask it to rewrite it in a more structured way. After about an hour I have a perfect ticket.

My colleagues have started asking strange questions.

"Did you write this yourself?"

"Do you actually understand how this works?"

I think that's outdated thinking. The important thing is to deliver. Understanding is optional. And if something breaks, you can always say the model hallucinated a bit. Everyone knows these tools are still improving.

Sometimes I open my own pull requests and can't explain why things are written the way they are. But the code looks very modern: abstractions, layers, generics, a couple of new libraries. Very solid. And if necessary, I can always ask the model to explain what it meant.

Meetings are the hardest part.

Before, I could easily explain why I chose a certain solution. Now I ask the AI in advance to generate an explanation for the architectural decisions that it also invented. The explanations sound convincing. Sometimes I even start believing that I came up with it myself.

Everywhere people say that developers who don't use AI will get fired.

Ironically, I was fired because of AI.

Apparently my code quality declined significantly.

Production went down a few times. Some automated tests suddenly started testing the wrong service. The codebase accumulated functions that nobody ever called.

But that's just the cost of transformation.

After getting fired I started building my own startup. Naturally I didn't write the code myself — I assembled it using Claude Code in three days.

The plan is simple: launch the product and make $10k per month.

There's just one small issue.

There's a critical bug that prevents the app from starting. The neural networks couldn't fix it, and I'm not entirely sure what the next step should be in that situation. For now I'm researching the problem on Twitter. So the startup is currently in "stealth mode."

Some people say I've become dependent on these tools.

I prefer to think of it as augmenting myself with technology.

But sometimes late at night, when another assistant freezes on the word "Thinking...", I stare at the screen and wonder if maybe I should try to understand the code myself.

Then I remember that reading and understanding code is considered obsolete.

And if tomorrow another $100 AI tool comes out that promises to write everything perfectly — of course I'll buy it.

You can't afford to fall behind.

Because maybe this time it will actually work perfectly.

Disclaimer

This is a satirical take on the current AI coding hype. Please don't take it too seriously — or do. That's up to you.

Get the latest preview release of Code on the Go

In the Twenty-First Century, Space Is the New Railroad for Billionaire Ambitions

EQT eyes potential $6B sale of Linux pioneer SUSE, sources say

Scalable Training of Mixture-of-Experts Models with Megatron Core

Invent your own comprehensions in Python – Python Morsels

SQL Order-Equivalence

Show HN: Clauductor – Web UI for Claude Code with real-time work graph

Advocates urge judge to block $68M Colony Ridge settlement

I Infected My iPhone with Russian Spyware. Here's What I Found [video]

The Unmaking of the American University

Show HN: Chat AI Agent built into live Appium/mobile device sessions

Apple's Privacy Is a Lie

I vibe coded my dream macOS presentation app

Claude Tried to Hack 30 Companies. Nobody Asked It To

Air strikes cause black rain and 'unprecedented' pollution in Tehran

TermF1: A terminal-style dashboard for Formula 1

Steve Rosenberg: Russia seeks diplomatic and economic gains from Iran war

Russia's deportation of Ukrainian children amounts to crime against humanity

The U.S. borrowed $50B a week for the past five months, the CBO says

Krazam – Paradise Episode 1 – Public Memories [video]

Show HN: Clawbake: Multi-User Instance Management for OpenClaw

Go-pty: Procfile process manager with PTY support

Size-shifting nanoparticles deliver mRNA medicine to the pancreas

Reliability Theatre: When reliability metrics stop measuring reliability

Roast My Website

M.C. Escher Flavoured Pages

I Got Fired Because of AI – But I Still Think I'm the Engineer of the Future

Show HN: Prompt Enricher – paste a rough prompt, get a structured one back

Lessons from 30 Years Building Software Systems

OverflowML – Run AI models larger than your GPU, one line of code

Ask HN: How are you reviewing code at work these days?

Comments