01/10/2026

AI is the new React (for recruiters)

Here it goes again

Following disapponting results in my entrepreneurial adventure, I have started to look for a job again. And I can't say I was prepared for the current job market.

Most offers I see these days, whatever the industry or company size, always involve genAI. Event when the role has nothing to do with it (my experience is in analytics / martech). That's okay, why not. After all, it goes without saying that only *real* companies ~do~ "MCP, RAG, custom harness, token maximization, familiar with langchain|graph..." all at once. Non-AI assisted coding is for "fucking losers", as DHH said recently, or, as I also saw it in random job description: "if you pride yourself in manually typing code, you should look elsewhere".

Keyword-stuffing job description with technologies acronym that recruiters don't know the meaning is the classic game and I've been used to it. Cheers to that. What breaks my nerve is a specific requirement that is so obviously ridiculous that it reminded me of the same, uninformed and stupid requirement in the early React developers job listings: the expected level of experience *must* be at least three years before the technology was released. True experts, as we all know, even pushed projects to prod before the technology even existed. Chuck Norris style. That's what drives me mad and that's what also signals the sheer, irrational panic from organizations these days. For instance, I spent a couple of hours looking for recent postings in my field. Reading through my screen, I kept seeing companies asking for 1/ demonstrated deployment of agentic workflow in production at scale 2/ Minimum 5 years of experience in agentic. It's so stupid and lazy that I am even considering the possibility that the company only wants openAI and Anthropic researchers. To work as SE or Head of data. For maybe a third of their salary, no RSU and actually spend their day in meetings. Yay... Sounds unlikely. But what's even more bothering is that the recruiters who wrote those description were lazy enough to skip the research part (i.e googling "What's a RAG?", "What's an MCP?", "Agent vs workflow") but proud enough to not use an LLM. Because the LLM would have pointed out the mistakes early. What an absurdity.

React, AI, the spin keeps spinning. AI has a twist though: companies are flexing tokens the Joan of Arc way, where burning is not a metric but the OKR. It does not feel like investing to me. Because if the use case was clear for every company, recruiters would even less actively look after the ideal, grand master before-agents-inception expert to get them replaced with an agent within 3 months. Being out of touch is the product.