The Great Coding Reset

Inside the months AI changed software engineering forever—and what comes next.

They are more productive than ever. They are deeply inspired. And some of them are absolutely terrified.

For decades, software engineers occupied one of tech’s most lucrative, untouchable, and high-demand roles. Today, they are watching their profession change irrevocably.

The shift hit like a lightning strike late last year. In the span of just a few weeks, OpenAI, Anthropic, and Google dropped next-generation AI models that radically overhauled their coding capabilities. Almost overnight, AI evolved from a basic autocomplete tool into an agent capable of executing complex architecture—the kind that used to take humans years of study to master.

The Point of No Return

Amy Surrett, an engineer based in Greenville, South Carolina, felt the ground shift beneath her in January. She booted up Anthropic’s Claude Code to build a sophisticated payment feature for a client.

"Coding a project like that by hand would have taken two or three days," Surrett said. "Claude did it in just over an hour. It felt like the point of no return. This industry is not going to be the same. My job is not going to be the same."

She isn't alone. Andrej Karpathy, a prominent AI researcher who recently joined Anthropic, noted in February that it was "hard to communicate how much programming has changed due to AI in the last 2 months." Before December, he observed, autonomous coding agents "basically didn't work." Then, suddenly, we hit escape velocity.

Flash forward to June 2026, and software engineering is undergoing a full-blown existential reckoning. The technology has sparked layoff anxieties, birthed a strange new subculture lexicon (like "tokenmaxxing"), and funneled hundreds of billions of dollars into AI infrastructure. Tech giants are leading the charge: at Google, AI now generates as much as 75% of the company's internal code.

AI-GENERATED CODE AT GOOGLE: 
[██████████████████████████████░░░░░░] 75%

Whether they like it or not, software engineers are Patient Zero in a massive global workplace experiment. Because coding operates on clearly defined rules and logic, it has proven far more vulnerable to AI disruption than other professions. The lessons developers are learning right now will serve as the blueprint for the rest of white-collar work as AI begins to automate other industries.

When Weeks of Work Evaporate

Software engineering has survived waves of disruption before. The desktop era gave way to the internet, which gave way to mobile. But the velocity of this current shift—supercharged by platforms like Lovable and Base44 that allow non-technical users to deploy apps without writing a single line of code—feels entirely different.

The sentiment among developers is encapsulated by a spin-off website launched by the popular engineering newsletter Latent Space in February. Its domain name says it all: wtfhappenend2025.com.

Case Study: The Video Encryption Crisis

The Developer: Kent Dodds, an independent instructor and former PayPal engineer.
The Task: Building a complex feature to allow students to download course videos securely via offline encryption.
The Old Way: Weeks of tedious, manual architecture and testing.
The AI Way: Dodds deployed an agent using Cursor (an AI-assisted development tool). It nailed the entire project on the first try.

"That was my first existential crisis," Dodds admitted. "I don't know what the ceiling is, or how fast we're going to hit it, but we certainly aren't anywhere close to it just yet."

'Agents Take Over the World'

At the AI Engineer Europe conference in London this past April, the atmosphere was a mix of techno-utopian hype and nervous energy.

"In the last six months, we have seen coding agents take over the world," declared Ryan Lopopolo of OpenAI's technical staff. He described a looming future where the role of the software engineer morphs entirely into that of an agent supervisor.

Traditional Developer: 
[Write Code] ➔ [Test] ➔ [Debug] ➔ [Deploy]

Modern Product Engineer: 
[Define Intent] ➔ [Supervise AI Agents] ➔ [Review & Refine]

However, this transition isn't without friction. Many developers express frustration over "vibe-coded" applications built by tech-novice colleagues, leaving professionals to clean up sloppy, unoptimized AI output. Others find their days disrupted, spent waiting around for token limits to reset or pivoting entirely toward writing hyper-specific prompts and specifications.

"It's basically not even worth my time to be manually writing code when I can have something like Claude doing it for me," said Danial Qureshi, a developer based in Toronto.

The Pivot to "Product Engineering"

If anyone can build an app with the click of a button, code itself becomes a commoditized utility. To survive, engineers are leaning heavily into uniquely human traits: empathy, context, and taste.

Dodds has completely overhauled his teaching curriculum. Instead of focusing on syntax and mechanics, he now teaches product engineering—prioritizing what to build over how to build it.

"I'm teaching the last skill that the last software engineer needs to have," Dodds said. The ultimate human moat is judgment: understanding what problems are actually worth solving and what will truly benefit the user.

The Evolution of the Day Job

For younger engineers like Surrett, who graduated in 2022 right before the launch of ChatGPT, the shift has rewritten the trajectory of her career.

AI Contribution (2025): 5% to 10% of her codebase.
AI Contribution (2026): 80% to 90% of her codebase.

"It's a double-edged sword," Surrett noted. "In some ways, I'm getting more done, but also doing less manual work, so it feels less productive." To keep her edge, she is sharpening her soft skills. "Figuring out how to phrase things, understanding what a client actually wants, having that human creativity."

The Big Picture: Jobs and Judgment

While tectonic shifts in tech spending have driven highly publicized corporate restructuring and layoffs, the sky isn't entirely falling. According to recent data from Indeed, software engineering job postings have actually seen a modest, steady uptick through early 2026. As AI produces mountains of code, companies are realizing they need human gatekeepers to ensure it doesn't break.

       [ Indeed Software Job Postings Index (2020–2026) ]
|                     ▲
|                    / \
|       ▲           /   \
|      / \         /     \       ▲ (Modest Uptick)
|     /   \_______/       \_____/
------------------------------------------
       2021   2022   2023    2024   2025  2026

Jason Young, a veteran engineer with 30 years of experience under his belt, isn't sweating the rise of the machines. Working as a lead engineer at ChargeItSpot, Young felt threatened by AI agents a couple of years ago. Today, he views them as mere tools.

For Young, the essence of the job has never been about typing characters into a text editor.

"The writing of text—that isn't what being a software engineer is," Young said. "Anyone who thinks otherwise has a wild misunderstanding of software engineering."