Welcome. If you're reading this you probably already know this, but my name is Radu. I’m a data scientist living in San Francisco trying to make sense of the speed run that is AI advancement. When I'm not making sense of it (very often), I make music, collect vinyl, and play outdoor sports like golf, pickleball, and tennis.
My weekends consist of frolicking around the city with friends, tinkering with Claude on new ideas, and attempting to find the best meals SF has to offer.
One thing I’m not doing on weekends or weekdays is writing.
And that’s the point.
Why I’m Writing
There are two distinct reasons I wanted to start this blog.
1. I Think I’m Losing My Edge
Recently, I realized something unsettling: my thought formulation feels slower. Not necessarily the formation itself but my ability to vocalize it and harden a thought into a clear, sensible perception.
During that realization, another one hit me: I haven’t touched pen to paper in years. Quite genuinely years. And almost everything I write electronically now, whether it be emails, code, documentation, is more than 90% AI-assisted.
I want to use this platform to rebuild my wordsmithery. To practice articulating ideas without relying on a model to smooth the edges for me.
Although I should make a disclaimer: the version of this piece you’re reading right now is definitely AI-polished (I swear I wrote the first draft myself). But that’s everything these days, isn’t it?
2. My Systems Weren’t Systeming
For a while, I thought my organizational skills were just bad. What I’ve realized instead is that my setup was inadequate for what I want to achieve.
I had:
Notes and checklists scattered across my Notes app
Daily check-ins with GPT (my therapist)
Health app for workout tracking
Partiful for social events
Teams for work
Google Calendar for everything else
My time-blocking was inconsistent. My weekly intentions rarely translated into execution. Everything technically existed... but nothing was unified.
I wanted one clean layout.
One source of truth.
One place where everything connected.
I wanted autonomy over my data. Over how my tools integrate. Over what those tools even are.
So I built it (with Claude at the wheel).
The Site
The public-facing version is this blog and homepage for all things Radu. The private side is my personal life agenda. This is a place to:
Create goals and long-term projects
Check in daily
Visualize progress with built-in data science tools
Take notes and capture ideas
Connect all of the above in one system
Mostly, I wanted a space where I could dictate new features as I thought of them.
Three years ago, this would’ve barely been possible for me. I might have stitched together a JavaScript template, fumbled through a buggy database integration, and shipped the ugliest frontend imaginable.
Now? Claude Code got it running in half a day. Minimal work from me beyond environment variables and buying a domain.
That shift matters. (Pretty good segway eh?)
The “Death of SaaS” and the Speed Run Era
Lately, headlines have been screaming about the fall of SaaS. Stocks plummet, rebound, then plummet again. With tools like Claude Code, OpenClaw, and hardware capable of running them, individuals can now build almost anything.
The future of AI is inherently speculative. In a space where the cutting edge determines who wins, narrative becomes as strategic as the technology itself. Every iteration ships with intentionally bold predictions that are part vision, part marketing, part defensive moat.
I think the market fallout is an overcorrection. The headlines are slightly overstated.
But the underlying fear? It’s not baseless.
There’s a real possibility that many SaaS products won’t perfectly fit individual workflows anymore. And when it’s this cheap and this fast to build custom software, investing in your own internal tools starts to make serious sense. There’s also an important nuance here. AI tools like ChatGPT are technically SaaS: cloud-hosted, subscription-based, centrally maintained. But they don’t behave like traditional SaaS. Salesforce gives you a CRM. Notion gives you documents. Figma gives you design tools. You conform to the structure they’ve built. Large language models, on the other hand, are more like cognitive infrastructure. They don’t just give you a product. Instead, they give you the ability to create one. That difference matters. SaaS optimizes distribution; AI optimizes creation. That’s where the tension lies, and also where the opportunity is.
As for software engineers who may or may not be reading this: our time has come. Either join the speed run or develop a skill AI can’t replace.
To be clear, I’m not even sure such a skill will exist forever.
But that’s a topic for another post, so I'll leave you with this.
My parting thought: take ownership of your workflow. Automation can amplify you, but it shouldn’t define you. Use these tools aggressively. Learn from them. Build with them. Just remember:
Autonomy isn’t the absence of automation. It’s control over how and when you use it.
What This Blog Will Be
This was mostly a welcome post. A slightly philosophical “about me” wrapped in a systems rant and a mild AI existential crisis.
But if you stick around, expect thoughts on:
AI and building in public
Data science
Music and creative process
Sports (get ready for my Bears takes) and hobbies
And whatever else feels relevant in this absurdly fast-moving era
I’m excited for what’s ahead—whether I get replaced by a bot or not.
— Radu