Over the holidays, many got their first real taste of the advanced capabilities Anthropic’s Opus 4.5—its most advanced AI model designed for developers—brings to the table when paired with Claude Code. They were building personal tools, small apps, half-formed ideas turned into working software, and in some cases, even ‘saving marriages’.
For some developers, this was the first time software felt accessible rather than gated by years of training and professional rituals. But for those who get a fulfilling high using programming to build apps, debug code, and find elegant solutions through logical thinking, this felt more like withdrawal.
Andrew D, founder of crypto tax software platform Awaken Tax, posted on X that even though he has a lot of fun coding, he can’t help but feel depressed. “The skill I spent 10,000s of hours getting good at. Programming…Is becoming a full commodity extremely quickly,” he lamented.
This is shortly after Google principal engineer, Jaaana Dogan, drew wide attention with a post describing how dramatically AI has changed coding over the past year. “We have been trying to build distributed agent orchestrators at Google since last year… I gave Claude Code a description of the problem, it generated what we built last year in an hour,” she posted on X.
Dogan later clarified, “It’s not perfect and I’m iterating on it but this is where we are right now,” adding, “The idea of everyone building software intimidates people.”
This comes just a few days after AI researcher Andrej Karpathy posted about feeling behind as a programmer, voicing a common sentiment among developers who feel that they can do a lot more, but they do not know where to start.
“I have a sense that I could be 10x more powerful if I just properly string together what has become available over the last ~year and a failure to claim the boost feels decidedly like a skill issue,” he said.
A strange gloom is descending on developer circles.
On one hand, they are shipping faster than ever before with tools like Lovable and Replit. Side projects that once took weekends now take evenings. Entire products appear in a single sitting.
And yet, behind the speed and the thrill, there is a quiet discomfort. The blazing speed at which AI is being embedded into software engineering is forcing developers to confront an uncomfortable question: What’s the value of their programming skill now that anyone can create software easily?
“Unfortunately, I find coding with agents to be a lot more productive but a lot less fun and interesting,” Charlie Marsh from Astral wrote on X.
Dopamine Crash
The anxiety is not really about jobs disappearing overnight but about identity.
A developer from a generative AI startup in Bengaluru told AIM anonymously that while using Claude Code has made it easier to build agents for automating coding tasks, speeding up his workflow, he felt a sense of loss over his programming skills. He admitted that AI tools are making him question his own skills, as now anyone can create software, even if with a lot of slop code or bugs.
Adithya S Kolavi, founder of AI lab CognitiveLabs, also agreed with the sentiment. “Productivity increase is really good. But the dopamine hit that I used to get by solving problems with code is completely gone,” he told AIM.
He said that while it’s fun to ship fast with Claude Code, the “satisfaction is gained by the ability to ship stuff, but not the actual writing code bit.”
What makes Claude Code especially destabilising is that it works. Not in demos. Not in toy examples. In real projects. And it is especially problematic for developers in India.
Threat to Indian Jobs
Anjney Midha, venture partner at Andreessen Horowitz, recently said at Moonshots Podcast that India’s IT sector, which is expected to contribute 10% to the GDP, according to some estimates, will get vapourised by tokens.
“If you’re India, for example, where double-digit percentages of your GDP are literally IT services, what do you do when Claude and GPT-5 tokenize vast portions of that flow?” he pondered.
IndiaAI CEO Abhishek Singh had also earlier flagged the threat of AI coding tools for Indian IT firms while speaking at the Bengaluru Tech Summit 2025.
Code is now cheap, fast, and generated at a rate that is completely out of developers’ and companies’ ability to read, understand, or maintain it line by line. “That alone breaks a lot of our old intuitions about code quality,” said a developer on X. He added, “If the end product behaves correctly, the internal codebase matters far less than we have been taught to believe.”
Others argue that the quality of AI-generated code will only improve over time. Andriy Burkov, the author of The Hundred-Page Machine Learning Book, and with experience in software development in India and Ukraine, said on X, “There was software development before Claude Code with Opus 4.5, and there is software development after.”
Not Everyone is on the Same Boat
McKay Wrigley, founder of Takeoff AI, summarised the developers’ depression on X: “I’ll observe that software engineers with 10+ years of experience seem to be the worst here because they’ve devoted a decade+ to their craft, and it’s hard to accept that AI is democratising the creation of software.”
Many developers built their sense of worth around craft. Clean abstractions. Elegant functions. Perfect architecture. Claude Code does not care about how proud you feel reading your own code. It cares about whether the system behaves.
Tory Green, co-founder of decentralized computing network ionet explained why that matters. “You didn’t spend 10,000 hours typing syntax. You spent it learning what breaks at scale, where reqs lie, how systems decay, and how to feel when something ‘works’ but isn’t actually right,” he noted on X.
However, Jean-Francois Puget, director and distinguished engineer at NVIDIA, disagreed. “You are most probably much more effective than a Claude user without SWE skills. What you build with it is probably much safer than what they build,” he posted on X.
Tory Green, co-founder of ionet explained why that matters. “you didn’t spend 10,000 hours typing syntax. you spent it learning what breaks at scale, where reqs lie, how systems decay, and how to feel when something ‘works’ but isn’t actually right.”
Green, though, argued that Claude Code only works “while someone in the loop still recognises wrong.” Claude Code does not eliminate developers. It strips them down to the part that cannot be automated yet.
The unease seems to be real. Once you remove the grind, you also remove the illusion that the grind was the point. With advanced tools like Claude Code and Opus 4.5, that grind seems to be firmly in the past.
The post Developers are in Existential Crisis, Thanks to Claude Code appeared first on Analytics India Magazine.
