Have we reached the age of personal software? Have we come to the moment that writing code is democratized, similar to how photography became available for everyone with the introduction of the smart phone? Currently I build what I (think) I need, or just what I like to build.
When I look up an app or a package, I first consider whether it covers no more features than my use case needs, otherwise I will build it together with Opus. That’s remarkable, as this thought contrasts a lot with how I used to look at software.
otherwise I will build it together with Opus
For years, the energy it took me to spend evenings after work on a hobby project wasn’t enough to bring most of my ideas to fruition. Either I got stuck at attempts or I made projects so small, mere prototypes, or with very few lines of code, just enough to reach to a visual (artistic) result: the effect of my code (what the code did) was downscaled due to the efforts it took me to build it.
In the age of coding agents, in no more than one year, being 2025, I worked through most of my ideas I had over the past 15 years. Still, I dropped projects because I lost interest, I refined and polished others, but I also finished up ones I cherished. Of course, I also initiated a lot of new ones. Why? For the fun of it. I like tinkering and building. Coding is still like Lego® to me.
The funny thing is that it seems that the AI companies have not expected software development to take such a leap. It feels like Anthropic seems to have created Claude Code almost by accident. A bit like Apple did not expect the Apple Watch to become a device that supports health and sports, marketing it as a fashion item initially. 2025 became the shape shift year of programming.
Will 2026 be the year of Personal Software? With the hype around OpenClaw one would expect so; that tool will fetch, arrange and build anything at the moment you need it.
We ended up in the age of YA, the age of Yet Another piece of software.
Ideas are the problem though; in all eagerness of both developers (addicted to the one arm token-bandit) and models (yes, the models are insanely eager to barf code), I see myself and others recreate variants of what’s already around: yet another note taking app, yet another coding agent, yet another ticket tracker, yet another speed reading tool. We ended up in the age of YA, the age of Yet Another piece of software. Yet another half baked vulnerable app hoping to get 180k stars on Github.
And I see the counter movement already: vibe coders tired of spending all those hours behind their Codex, AMP or CC. Where’s that fatigue coming from? Staying up too late? Or have we drawn the conclusion that all those hours did not bring the product we hoped for? Is the real world of SWE harsher than we believed it is, blinded by the potential of vibe coding? Or are we getting to understand that software engineering, systems architecture, code organization and data flows are still problems that we as humans need to fix? Which we just aren’t good enough at? That we’re not as senior as our role names might have suggested?
We might need to take a little step back and reflect on what we need and what we want and how we want it to be. Delivering value is not easy. Delivering value securely is harder. Delivering value user friendly and securely is even harder. Let’s take time to plan, architect and deliver value, and let it be built by agents, because they’re very good at exactly that.