In the last several months I’ve been buildin’… software. Since my childhood years, I’ve always had the problem of remembering lessons and subjects 1:1. Despite the fact that I was in front of computers from the age of 9, programming was never ever my thing - due to the reasons already mentioned, I’m really bad at boilerplate. My line of thinking is always logical and the moment I face abstraction I am simply off, maybe it is some form of ADHD (we millennials didn’t have such modern things). I can read code, I can understand what a program is doing - this is the base of my career in software testing for the past 20 years, but writing code is the single computer thing that I really despise.
So when LLMs came to the party and became ridiculously good in the last few generations of frontier models, man… it was a fuckin blessing. From automation to writing small solutions and now building a complete self-healing automation system that is actually useful.
For the past two months I’ve been working on two solutions to two of my actual problems:
- A photography library application in which you can find a photo by just putting in your memories (photofindry.com)
- The simplest secure solution to connect two damn computers over the internet, via the infrastructure you already have to host your website - a VPS (OK, maybe not everyone has a VPS). OwnTransit - my first full-fledged open source project, as every security-related project should be in the age of AI. https://github.com/sentrybottale/OwnTransit
$ ls ./solutions
Select a project to inspect it.
Now, did I build software that everyone will find useful? That is a completely different story, especially in the age of AI and social media, where literally everyone can do something, but nobody will know - since that is not interesting to the social media algorithms.
While building these, I see the big threat: everyone can build a software solution, and if you are the one selling software solutions, then the only way for you is to build even more cool stuff. Failing to do that is a death sentence, especially for these big behemoths of the past, some of which I worked for. And it is huge pressure! There is no longer a way in this world to sell someone a 50 million EUR project, absolutely no way! And even the big tech services are at risk - I successfully killed my Google Photos subscription and my remote computer access solution. Both are built by great companies, with great people, but that doesn’t mean anything now.
$ trace ./industry
Build
Everyone can build a software solution.
Pressure
Companies selling software solutions have to keep building more.
Customers
The ones who looked after their customers can make it through the storm.
Select a theme to follow the argument.
And so many people in these companies are losing their jobs not because of AI, but due to the software industry collapse under its own pressure. It is an “asteroid” extinction event, where every million-dollar company will be put to the ultimate test, and all the neglect of their customers over the years, every bad managerial decision, will be paid, traditionally, with the livelihoods of so many good people working in these companies. We can’t do much, that’s why it is so sad, but it is a process that has a beginning and an end. And I’m an optimist - in the long run it will settle down, and the only winners will be those that were looking after their customers before the storm, during it, and after. This is the only thing that is working!