The last post here is dated April 26. It’s July now. That’s not a break, that’s a disappearance.
I could pretend it was intentional - a deliberate step back, time to think, all that. It wasn’t. I just ran out of the specific kind of energy that writing requires, and I ran out of it in three different directions at once. A new job, the tail end of a master’s semester, and a business idea that ate more evenings than it deserved.
Here’s what actually happened.
May: A New Job, and Being a Beginner Again
I changed jobs in May.
There’s a particular feeling that comes with joining a new project, and nobody warns you about it properly. You come in with years of experience. You know your patterns, you know your language, you’ve seen enough codebases to have opinions. And then you open the repository and understand approximately nothing.
Not the code itself - the code is usually readable. What you don’t understand is why. Why this service talks to that one through a queue instead of directly. Why there are two models that look like duplicates but aren’t. Why that one module everyone tells you not to touch exists at all. Every codebase is a fossil record of decisions, and you arrive without the context that makes any of them make sense.
So May and June went into rebuilding that context:
- Reading before writing - resisting the urge to “fix” things I didn’t understand yet
- Asking a lot of questions - and writing down the answers, because I’d forget half of them by the next day
- Small changes first - the boring tickets are the best map of a system you’ll ever get
- Learning the domain, not just the stack - the business rules were harder than the framework
That’s a full-time cognitive load. At the end of a day spent building a mental model of somebody else’s system, sitting down to write a coherent blog post is not something my brain was willing to do.
Meanwhile: The Semester Wouldn’t Wait
At the same time, I was finishing another semester of my master’s in computer science.
The timing was, let’s say, unfortunate. Onboarding at a new job and closing out a semester are both “everything is due at once” activities, and they don’t take turns. Projects, deadlines, the whole thing - all landing in exactly the weeks when I had the least room for it.
I got through it. But something had to give, and it turned out to be this blog. I don’t regret the prioritization. I do regret that the writing was the first thing to fall off, because writing is how I actually process what I learn. Skipping it means a lot of May and June is stored in my head as a vague blur rather than something I can point at.
The Digital Agency Idea (And Why I Killed It)
Somewhere in the middle of all this, I got the idea to start my own digital agency.
It’s a natural thought for a developer. You have the skills, you’ve seen how projects are run, you know what good delivery looks like. Why keep building someone else’s product when you could sell that capability directly? So I did what I always do before committing to anything: I researched it properly. Weeks of it. Market size, positioning, pricing, what actual agencies in this space charge and how they find clients.
The research killed the idea, and honestly it did me a favor.
Two things came out of it:
- The market is saturated beyond reason - not “competitive,” saturated. Every niche I looked at already had established players with portfolios, references, and years of relationships I couldn’t shortcut.
- AI tools have collapsed the floor - the work that used to be an agency’s bread and butter (landing pages, simple integrations, basic CRUD applications) is now something a competent non-developer can get to 80% with AI assistance. The entry-level tier of the market is largely gone, and that tier is exactly how you’d bootstrap without capital or an existing network.
Put those together and the honest conclusion was: starting from zero, with no client base and no differentiation, this isn’t a business - it’s freelancing with extra overhead and worse margins.
It stung a bit. I’d gotten attached to the idea. But finding out in the research phase rather than eighteen months and a company registration later is the best possible outcome. Deep research is cheap. Learning the same lesson by living it is not.
So Now: Looking at SaaS Ideas, Slowly
Which brings me to where I actually am.
I’ve moved on to analyzing SaaS ideas instead, and I’m being deliberately unhurried about it. The agency research taught me something worth keeping: the hard part isn’t the building, it’s picking something worth building. I can ship a product. That was never the constraint. The constraint is finding a problem specific enough that AI-assisted commoditization doesn’t flatten it in a year.
So the criteria I’m filtering ideas through now:
- Narrow beats broad - a boring problem for a specific audience beats a general tool with a huge addressable market I can’t reach
- Domain knowledge as a moat - something where knowing the field matters more than knowing the framework
- Reachable customers - if I can’t describe exactly how I’d find the first ten users, it doesn’t count as an idea yet
- Not trivially replicable - if a well-prompted afternoon reproduces the core of it, that’s not a product
This is slow work. Most of it is reading, talking to people, and discarding things. There’s not much to show and even less to publish, which is part of why the blog stayed quiet - “still thinking about it” makes for a bad post.
Was the Break Worth It?
Partly.
The job change and the semester weren’t optional, and I’d make the same call again. The agency research was genuinely valuable - I got a clear “no” for the price of some evenings, which is the cheapest “no” available.
What wasn’t worth it was going completely silent. Writing here was never supposed to be a series of finished conclusions. It was supposed to be a record of what I’m working through, and the last three months were dense with exactly that kind of material - onboarding, research, an idea dying in an interesting way. All of it worth writing about, none of it written down.
So this is me starting again, with a lower bar. Posts about things in progress. Posts about ideas that didn’t work. Posts about what I’m learning in a codebase I’m still figuring out.
The SaaS thing is still cooking. When it turns into something more than a spreadsheet of rejected ideas, that’ll be a post too.
Just probably not in three months.