Almost 20 years ago, I asked a computer scientist friend for advice about how to get better at using computers. He recommended that I brush up on sed, awk and regular expressions.
Since then, I’ve used modest scripting skills to create content that I post on my own website. The website is written, by hand, in static html and looks like it was written in 1993. But the content is more important to me than the presentation, and I spend most of my time related to the website adding and updating the educational videos I create and post.
Now, it is 2026, and everyone I know with any computer credentials…is encouraging me to vibe code an update to my website. What is interesting, is that those same people pat themselves on the back for their “smart” implementation of AI, which is based on leveraging all their canonical knowledge of computer science. In other words, they “paid their dues”, and that makes them, supposedly, better AI adopters. But they are not telling me to pay my dues.
Anyway, I am not taking these people’s advice, because: 1. I know that’ll result in a dependency on AI, 2. No one can seem to articulate what should be done to the website; rather they just speak vaguely about what is possible, 3. My selfishness is more geared around my own understanding of a process, rather than a certain outcome, 4. I prefer slow, incremental, transparent changes to my workflow, rather than something that is incomprehensible.
This is really cool! I made a home studio planner with Claude that handles my pachbay planning and make it easy to rearrange it when I change stuff and can also optimise it based on length of the patch cables k prefer to use Also made a Jaspers stand planner which is nice to rearrange gear to get ideas on how to execute.
Simple nice things to have that I think vibe coding is amazing for. Will definitely check out you studio app
I don’t even remember now. That was in June, which feels like 5 years ago in terms of AI. I am sure there is something leaps and bounds better now anyway.
E: I’m only a semi hater on this stuff. Yall do you.
I didn’t come in here to bash but these apps, utilities and firmware “reimaginings” have definitely invaded many of the spaces I used to go for information on DIY effects and synth building.
Avalanche feels like an apt description from my vantage point.
This itself sums up perfectly how malignant and banal all this stuff is to me. Meh, it was a few months ago so who cares about all those resources that were used up, lets just use up a bunch more to get another iteration of it.
I cannot fathom what it must be like to be an actual company/dev/regular human that needs water these days.
The slop and the hype and the insane economic manipulation and the deranged resource consumption is all real and all very troubling, to say nothing of how the reptile class is applying the technology. This (reasonably and understandably) masks many of the things that AI is genuinely great for.
In a responsible and mature professional environment, I think that now AI can and regularly does deliver a genuine 2x-3x of high-quality output by high-quality engineers - I believe that’s an accurate description of my team, and that’s about what we’ve seen. Hilariously, this sounds very disappointing to business psychopaths, but two years ago no amount of money could magically buy you a permanent 2x output at the cost of a few months of onboarding a new technology, in terms of best practices and engineering culture.
I also find that, when used with a modicum of common sense and skill, it’s an unparalleled tool for discovery and domain exploration. As long as you don’t just lock yourself in a research closet with Claude and your farts, and you stress test the things that you’re “learning” from your conversations, you can spin up on a new technology or field at a speed that would have been hard to imagine a couple of years ago. It’s not “just Google on steroids” in the same way that the cloud is not “just someone else’s computer” - it’s a different, significantly better thing.
Unfortunately, the costs that the market has decided that we should all pay for this are far too high - certainly in my opinion, anyway. And I don’t get the impression that most of the code that’s being produced in professional contexts is being produced in a mature and professional manner. I wouldn’t be sad at all if it just went away tomorrow, but of course that’s not going to happen, so I have to learn to live intimately with it if I don’t want to figure out another career. I’m still trying to figure out the ethical implications of that decision.
The percentage of new ‘tools’ my colleages vibe code that simply do not deploy successfully the first (second, or third) time on a real staging VM is close to 100%. Dude, just because your app works on localhost:15969 does not mean anything…
Some of them eventually get there, but these days I make them complete an internal course in Docker and Git fundamentals before I will spin up a staging machine.
I’m really struggling with the security side of it, because I’m the ops guy, so it’s kinda-sorta my responsibility to care about it (it’s everyone’s, but you know). My new pithy aphorism that my coworkers are probably getting tired of hearing is “if you built a sandbox for your agents and you haven’t instructed them to try to break out of it, you haven’t built a sandbox for your agents.” It’s a sobering experiment.
This feels like a great place to share that I work on several plugins in juce and also some html tools to improve workflows and reduce the time and effort of studio work.
But also I have mixed feelings about posting this stuff here because I see how polarizing the topic is.
Also made a Jaspers stand planner which is nice to rearrange gear to get ideas on how to execute.
