Raw LLM Responses

I'm a senior developer, been in the industry for 12 years now and I was always somewhere around hard problems and state of the art tech. I've been using LLMs to help me in my work since first appearance of chatgpt - even that primitive model was helpful to me. I could give it small chunks of work and lots of explanations, but I felt like this was still faster to integrate these pieces than writing everything from scratch. Since then I can't beleive how much my workflow changed. Today I mostly talk to agents, do a lot of explaining, lots of thinking, orchestrating, directing and verifying. But very little namual coding anymore. Also my skills suddenly became easily transferrable and language agnostic, as much as I thought this certianly won't be the case I admit I could be writing code in any language with the help of AI agents now, even those I never used. Maybe except something really cryptic like Erlang. The amount of work that still commands humans writing code is diminishing drastically. My productivity is through the roof now. Now we're at a point where I'm actually loving it - I love doing the thinking, designing, suggesting and analysisng solutions and now I get to do that most of the time. And yet I'm looking towards the future worried - how long before these models would be doing all this work in my stead? Today this actualyl feels far, but I can see it slowly nearing. Then what, whole industry would be only founders chatting with agents on their own? All this progress took only about 3years, from great tool to everyday industry standard and major enabler. I'm a bit scared to look next 3 years into the future of my industry.