Stop Begging Billionaires To Fix Software — Build Your Own
from the take-control-for-yourself dept
This is part one of a two-part series on using AI tools as one piece to fight back against tech company control of our lives. Part two shares the details of how I built my own task management tool in a few days for next to no money and without any coding skills. But first, we need to talk about why this matters.
Earlier this year, I wrote about how people need to take back some control and agency over the tools they use online. As I wrote at the time:
The internet was supposed to liberate us. Instead, it’s left us feeling helpless, waiting for billionaires, governments, and tech giants to save us.
The most insidious thing about Big Tech’s takeover of the internet isn’t the concentration of power — it’s how it’s trained us to beg for scraps from our digital overlords.
So I decided to practice what I preach: I built myself a sophisticated personal task management tool using only AI — no coding knowledge required. Within three days, I was using it to manage my workday — and I keep making it better.
Here’s the thing: I know many of you think AI is overhyped bullshit. That it’s just another way for tech companies to extract value while delivering nothing useful. And you’re not entirely wrong about some of the hype. But, at times, those same AI tools can actually help you escape the very companies peddling them.
You can actually use those big tech company AI tools to escape some of “Big Tech’s” hold on your digital lives.
I now have a tool that works exactly to my specifications. I control the code, the hosting, the data. No company can enshittify it. No growth hacker can “optimize” it to extract more engagement. No billionaire can wake up one day and decide to charge for verification badges, import Nazis, or kill third-party apps.
It’s personal. It’s intentional — in that it’s based on my own intentions directly, rather than forcing me to fight a service provider’s best interests when they conflict with my own.
This is what “vibe coding” offers: the ability to build personal tools without knowing how to code, using natural language to create exactly what you need.
It’s not software from some big company trying to upsell me, grab my data, or control me.
In my original post earlier this year, I pointed out that we now have a generation of internet users who grew up entirely on giant platforms controlled by billionaires who believe their only recourse for anything is begging those same billionaires — or the government — to fix things.
But the nature of an open internet is that you have agency and you can make some choices yourself, taking back the power for yourself, rather than demanding that others make them for you.
The early internet was about many people building things themselves, often for themselves. The rise of giant platforms trained us to accept their constraints as immutable facts of digital life. But they’re not.
Yes, there are legitimate concerns about AI, including about the companies that provide it and how they operate. But dismissing the technology entirely means missing how it can be weaponized against the very systems of control we’re trying to escape.
Let me be clear about what this is and isn’t good for:
- Do use it for personal productivity tools, workflow automation, and small community projects
- Do think of it as digital DIY — sometimes you need a professional, sometimes you can handle it yourself
- Don’t use pure vibe coding for anything involving serious security, payments, or legal liability
- Don’t build public-facing apps expecting millions of users — at least not unless you’re okay dealing with major problems down the road
For social platforms, protocols like Bluesky’s AT Protocol are key (disclosure: I’m on the board). But for personal tools that don’t need social? Vibe coding changes the game entirely.
It was only in early February of this year that Andrej Karpathy coined the term “vibe coding” to describe using AI tools to code things for you without needing to worry about all the details. You literally just chat with an AI until you get working software.

The concept exploded. By now, just four months later, there are already at least 15 books on Amazon about it (though the only serious ones — by Steve Yegge/Gene Kim and Addy Osmani — aren’t out until Fall, and I assume many of the others are AI generated slop).
This isn’t just “AI-assisted coding” where you still need to understand programming. This is describing what you want in plain English and iterating until it works. No coding knowledge required — my own coding skills are basically non-existent as the last time I really did anything directly with code was in the 1990s.
It turns out to be quite powerful. And liberating.
Ernie Smith used vibe coding to improve his own workflow, building a tool to use Obisidian (the popular note taking app) to post directly to his blog, which is built on Craft CMS. Smith, who is not prone to hyping AI tools notes how this seems like the proper use of the tool:
This is a situation where LLMs helped me solve a “me” problem without getting in the way of anyone else. I’m not going to be using LLM-created copy or images. But I did just figure out a way to save myself a ton of time when uploading a post, which I hope will make it easier to do so over time.
That’s exactly it. This isn’t about creating the next unicorn startup or replacing human creativity. It’s about solving personal friction points without waiting for some company to maybe add the feature you need (right before they pivot to crypto or whatever).
My task management system cost almost nothing to build and runs exactly how I want it to. No subscription fees, no feature creep (unless I decide to add those features), no suddenly discovering that my workflow has been “optimized” for someone else’s engagement metrics. No chance of waking up and finding out that my tool is owned by a fascist.
This is what taking back control actually looks like. Not begging Zuckerberg to “do better.” Not yelling at people on social media. Not hoping the government will save us. Just building what we need, for ourselves, on our own terms.
The companion piece shows you exactly how I did it for myself. But the why should already be clear: because what good is the open internet if we’re not using it for our own purposes.
Filed Under: agency, big tech, control, decentralization, enshittification, vibe coding


Comments on “Stop Begging Billionaires To Fix Software — Build Your Own”
This has always been the answer. Before AI-generated “vibe coding” it was just coding; you picked up a book (earlier) or read an online tutorial (later) on COBOL or javascript or whatever and you just did it. As time marched on the barrier to entry (the amount of learning you have to do and/or structural systems you need to set up) has decreased massively, and that’s great.
The digital prison some people find themselves has always been self-imposed to some degree. The bars have only gotten softer and the gaps between them wider with every passing year. It’s a bit perplexing for anyone to say they feel trapped by these things in 2025.
What's this?
Techdirt community members with Techdirt Credits can spotlight a comment as either the "First Word" or "Last Word" on a particular comment thread. Credits can be purchased at the Techdirt Insider Shop »
…
Until you find yourself sued because your code is 1 to 1 stolen from corporate software.
The other issue of course is that llms are trained by the large corporations. It won’t be long until they figure out how to push answers towards their own tools and products.
Re:
That would be like an LLM perfectly recreating a specific literary work based on general story prompts. That’s not going to happen unless you set out to make a 1-to-1 copy of an app, and even then, I doubt it.
Google tried something like that with their Shopping service. They were fined $2.7bn by the EU for it.
Re: Re:
Except large chunks of corporate code are also ai created. I think you underestimate the ability an llm to spit out code verbatim. It’s one of the reasons that open source projects are blocking and refusing to use ai code. Beyond copyright, code is licensed for particular uses.
And? Hoping and trusting governments to act doesn’t stop it from happening for the years it usually has to happen for.
Re: Re: Re:
Yeah; it seems to me that the issue isn’t “hey, you copied those chunks of UNIX SVR4 into ACOS” but more like what happens on YouTube: “Hey, this code in our enterprise product looks exactly like that code in that open source project! And our enterprise product is 2 years older! Let’s SUE!”
…and after discovery, they realize the same AI generated both sets of code, and only the lawyers and their AIs get rich.
Re:
If it’s not getting sold, how are they going to get a copy of it?
Re:
How the fuck are they going to sue you for code you only use personally?
I’m looking forward to your follow-up to get practical guidance.
In particular, I’d really like to see a layperson’s explanation of how to connect what you write in ChatGPT (or whatever LLM) to the software you want ChatGPT to manipulate.
I also love the framing about not having to beg big tech companies to do X. That’s basically 99% of platform governance research and advocacy, and I’m so over it.
Re:
You give it the ability to write a message to a tool or to read from a tool. Basically instead of reading from your message or from its stored data it triggers from a tool it’s been given access to to receive data. Or instead of sending a response back to you it’s sent to a tool instead that can execute it or take action based on it.
However I would use extreme caution when using AI to edit or create things like data without full human review. As the below article demonstrates AI makes stuff up and that can happen even when it did not before. So if you need a reliable, repeatable and precise system it should be a program or code itself that the AI can execute, not AI done directly.
https://www.techdirt.com/2025/05/21/whoops-chicago-sun-times-publishes-ai-generated-summer-guide-full-of-made-up-recommended-books-nonexistent-people/
This has always been the answer. Before AI-generated “vibe coding” it was just coding; you picked up a book (earlier) or read an online tutorial (later) on COBOL or javascript or whatever and you just did it. As time marched on the barrier to entry (the amount of learning you have to do and/or structural systems you need to set up) has decreased massively, and that’s great.
The digital prison some people find themselves has always been self-imposed to some degree. The bars have only gotten softer and the gaps between them wider with every passing year. It’s a bit perplexing for anyone to say they feel trapped by these things in 2025.
This is pretty cool. There’s only a few caveats. The big one for me:
In your other post, it looks like you’ve created a log in session, and some stuff is hosted in the cloud. It’s worth keeping in mind that this is now an attack vector.
I think one thing you’re missing is that basically all of those problems are things like social networks, which come with coordination problems or moats.
This democratizes things to a new level, but the barrier to entry was already low enough for most average software. People can and do whip up things like Shazam clones pretty easily. Really what this frees you from is the petit bourgeois, not the Zuckerbergs, if you don’t have to worry about getting monetized via ads or data selling.
It’s going to be both, unfortunately.
(It also does make me a little sad, that more people will never actually learn to code. There is a beauty and structure to it that you miss with vibe coding. But I suppose that is the trade off for accessibility, especially for people who were never going to learn it to begin with.)
Re: To be fair
Most people will also never learn to fix problems with their cars, and they’re perfectly happy taking it in to the shop when something breaks.
Several months ago, my AC blower died. Here in the South, that’s nearly a death sentence. I found the part at a local parts store for a reasonable price, then watched a few videos to see how to switch it out (RIP Haynes) and got it done.
It wasn’t hard and I was happy I learned something new while saving a bunch of $$$.
As a software developer I have long rejected non-open source software, and fixed the problems I encountered (though constrained to the effort to fix it vs my resources).
I encourage everyone to meaningfully have “agency” over your software. Learning to code it always a useful skill in my opinion. But even if you don’t, you can have others (people you know/trust, or hire or w/e… it’s your responsibility) help you. And and Mike is pointing out, there are lots of other tools as well.
I did something similar when I built my news site back in 2013.
I approached local newspapers and other local news organizations saying that I have a major background in technology, internet, and digital rights, so I can add a huge amount of value to the news organization with my coverage.
The TV news organization said that people aren’t interested in the internet because the internet was just a silly little fad that no one cares about (yes, that was in the early 2010’s). The newspaper told me to “come back when you right real news”. Every other news outlet ignored my pitch of generating news articles related to the internet and technology. The general sense I ended up getting was that the internet was unimportant to them and no one cares about anything related to the internet and technology.
So, I built my own site and continued my news writing career after the previous website owners I worked under called it quits. My site rocketed to become the number 2 news website in my entire city. I was beating out whole news rooms by myself which was, not going to lie, ego boosting.
When I returned to these news organizations to show that there was interest, I was told “you cheated” before explaining that I was too stupid to pull something like that off. This before they immediately ended the conversation on the spot. So, I said “screw ’em all” and kept working on my site. One of those local newspapers was already forced to declare bankruptcy after. The others have been on the ropes for years now. I have little sympathy for them after the way they treated me.
"AI"
My girlfriend and I call it “AC”
Artificial Cleverness.
What we have created is by no means intelligent, but certainly humans are quite clever.
I have not seen any evidence in fiction or reality of anything resembling the creation towards intelligence.