I ride the AI-hype train

I've been writing code for more than a decade. Occasionally, I'd stumble upon interesting challenges at work, however, most of the work I've done is putting forms, buttons, tables, APIs to handle that, writing deployment scripts, etc. Omg, we're using Golang now... a week later ... boring. 95% of work (totally made up number) becomes a mundane repetitive patterns. "Oh geez, I'll try this cool thingy..." -> a week later back to boring.

I built a vite-plugin that took json definitions and generated forms with react-hook-form, shadcn, zod, etc. It supports multi-step forms, validations, etc, etc, etc. Produces a really nice and responsive UI. And it's responsive not with boring screen widht, but with a container query! Ha! It was interesting, it was "somewhat" (by somewhat I mean more like understanding vite plugin API) challenging... anyhow, it was interesting. It was out of ordinary and increased a performance of our app with codegeneration for some critical UI features.

However, I had to come up with some BS excuse to explain why I'm spending time on that...it made my team's life easier. Reality is, we could have been writing forms as usual, connecting them to API. I've helped everyone to understand how we can make write better forms. Also built a plugin that abstracted the complexity. Frankly, we didn't need to abstract that. But it was cool. I choose cool thing over boring thing – I suck as a technical decision maker.

I implemented a very cool client-first state management approach for our web apps, a combination of rxjs + indexeddb + a concept called "Streaming sync collection". Works perfectly amazingly fine, I can literally shovel ~70k records on the client and filter/order/map there. User's think the app is so fast that they don't believe it's actually a website. I did set-up the framework and documentation... but then, well, now it's dull work to bring over all of the collections.

I've done a lot of backend architecture, DevOps... I can't believe that I'm saying that, but building a bespoke optimized UI experience somehow (to me) feels more interesting than most of DevOps tasks.

I've recently built a script to produce production containers with a 100 lines of bash, just so I can go home and spend an hour schizoranting to my girlfriend, why "Earthly", "buildx bake", et cetera sucks... and how the modern DevOps is not real and they're all weak. "Back on our days, we ....". I sound like my grandparents, who walked 35km to school in -45C, with snow to their necks. Anyhow, my girlfriend seemed bored.

The reality for most of us is that we're building these dull forms, pages and APIs (omg, we're using fucking GraphQL, or pure RESTful, or it uses this new state management or whatever other dumb exuses we come up with to make us feel that what we do is interesting).

Reality, is that most of us are not ThePrimeagen or Theo or Linus or whoever other cool person. We're mostly doing boring stuff. I am genuinely happy for you if you entertained by the work you're doing. I am bored with boringly-repetitive web development (or full stack, or web master, or whatever you call it nowadays).

My boss literally couldn't give a flying fuck if I write the code or if it was AI. My boss is business. Business cares about outcomes, not about "omg biome lints everything so much faster now". He only cares that the thing works and he can sell it. Obviously, we do have constraints that we need to pay attention to, e.g. security/compliance. But I don't care where the code comes from as long as I can confidently think that it's good.

I do have a shitload of ideas for cool side projects, but, I rarely have time to work on them. I spend most of my time at work.

I need to relaunch a product now. We have built similar product before. We already have business processes defined on paper. We understand what needs to be done. I've done it before. I don't want to do it again.

I vibe-coded the shit out of it in four days, reviewed the code. I think the code is good. I tested it and understood it. It works. It solves the problem. I have fun pentesting it.

Also, I finally have time to work on that cool idea, and write it in Zig. I can work on my cool stuff myself, without AI. I finally have enough energy to develop my skills and work on things I am genuinely interested in.

AI can do boring stuff, create those forms and pages all day long. I don't care. I can do cool stuff now.

Will AI eventually make people dumb? Yes, 100%. No doubt. But, now – at least for a while – we can use this extra free time to actually spend time on "ourselves". Eventually, corpocucks, will demand that we spend 70 hours a week generating AI slop code.

I guess I'll enjoy the freedom while it lasts.