The frontend is dead, long live the frontend! Evolve or die. What? Let's figure it out.

I've been in this business for a long time: I saw the era of BBS's and the sunset of perl-based sites, I saw the rise of php3 and jQuery, and how React killed everything else. And remember the profession of "HTML/CSS coder" or "Markup developer" (aka "verstalshik")? Well, it looks like classic frontend developers will soon follow them.

And it's not just that Claude can finally vertically align an element and you no longer need a senior dev with 15 years of experience for that - it's about the value of frontend as a whole.

🍄 Point 1: damn chats everywhere

Frontend is the UI for customers, which gives them the UX they pay for. Frontend devs make the frontend.

But what happens if all UIs get replaced by chats with a couple of buttons? - We'll get unification.

Literally, right now you can build a fairly complex startup where there's nothing but a landing page and a button to connect an MCP/skill to your Claude Desktop and ChatGPT. Moreover, sometimes a landing page isn't even needed, because everything is already inside agents ecosystems.

Somewhat reminds of the Telegram bots story, but at x1000 scale.

Why do we need frontend devs if there's no frontend?

🍄 Point 2: disposability

For the past week I've been working on a project, running 1-3 hour AI loops in Claude Code and Codex multiple times per day, and of course the resulting code was complete crap.
More precisely, overall it was more like the miracle of the creation of the world, but if you zoom it in places it was total garbage.

But that doesn't matter, for two reasons:
- I could point the agent at any place in the project, show it the browser (mcp devtools is love), write "this s**t doesn't work here, fix it", and it would fix everything
- despite extensive ideation, I missed the proper data flow, but even big logical misses were eliminated in half an hour with another loop run

So my value as a frontend developer was even lower than my value as a manual tester. And the entire frontend was literally disposable, and whole chunks were redone at my command in literally half an hour per case.

Now imagine an agent that watches how user clicks on a site, analyzes it, and then adjusts the UI itself to make the UX better for users. Almost live. Sounds crazy? Let's discuss it in 5 years.

🍄 The survivors

Meanwhile, the value of backend and cloud engineers is still much higher, due to questions of security and resources.

Although, I saw with my own eyes how Claude Code clicked around in my browser in the Google Cloud Console - it was really scary, though in the end everything went off successfully.

So what to do? At least N years from now.
- developers - accept fate and develop soft skills, primarily around product management and agent orchestration
- companies - look for these soft skills, look deeply into ways of working, not just at whether a person knows something about AI and Claude Code (which will be obsolete again in half a year), plus traditional non-coding skills
- another option - go work in chop-shops, where frontend wizards with long beards and 30 years of experience hand-carve Fabergé frontends, while others slather it on with a thick brush of agents, earning millions - there will always be demand for high art and its own clientele

Or am I being too gloomy? What do you think about the future of frontend developers as a standalone profession?

#Frontend #AI #Coding #SoftwareDevelopment #FrontendDevelopment #WebDevelopment #ClaudeAI #AIAgents #FutureOfWork #DeveloperTools

Original post: https://www.linkedin.com/posts/alexander-remi_frontend-ai-coding-activity-7453310522881359872-EHZZ