Should You Install It—or Just Build It?
When the cost of small, targeted custom code drops, the rational answer to “plugin or build?” changes. A practical look at fewer dependencies, less overhead, and when generic software still wins.
Should You Install It—or Just Build It?
You know that moment when you open a drawer looking for one simple tool—and end up digging through a pile of gadgets you barely remember buying? I had that recently trying to find a screwdriver. Turns out I owned five multi-tools, none of which actually did the job well. Software systems tend to grow in exactly the same way.
What started as “we just need a small feature” slowly turns into a stack of plugins, tools, and integrations. It works… until it doesn’t. And by the time something needs to change, you’re no longer dealing with one simple requirement—you’re negotiating with an entire ecosystem.
That’s really what this piece is about. Not just that AI is changing software development (you’ve heard that already), but something more practical:
It’s becoming viable again to ask: should we install this… or just build it?
For years, the answer was obvious
If you needed a feature—a form with logic, an API integration, a workflow—you didn’t build it. You looked for a plugin or a SaaS tool. Custom development was expensive, so you accepted software that did almost what you needed and worked around the rest.
And honestly, that was the right call.
That’s how we ended up with ecosystems like WordPress: incredibly flexible, powered by themes, plugins, and integrations. You could build almost anything by combining the right pieces.
The quiet trade-off
But there’s always been a quiet trade-off.
Most software gets you about 80% of the way there. The remaining 20%? That’s where things get… creative. Extra plugins. Custom scripts. Little workarounds layered on top of each other.
Individually, they’re harmless. Together, they add up.
Until one day:
- an update breaks something
- a plugin is no longer maintained
- performance starts to dip
- or a “small change” turns into a risky operation
At that point, you realize something slightly uncomfortable:
You’re carrying a lot of software you don’t actually need.
How AI subtly changes the equation
Now here’s where AI subtly changes the equation.
Not by replacing developers. Not by magically building full systems.
But by making small, targeted pieces of custom software much cheaper to create.
The kind of things that used to feel unjustifiable:
- a custom validation rule
- a lightweight admin tool
- a simple integration
- a bit of business logic specific to your process
These used to take enough time that installing a plugin—even an oversized one—was the rational choice.
Now? That balance is shifting.
You can often build exactly what you need, without dragging in everything you don’t.
What I’m seeing in practice
I’ve been seeing this more and more in practice.
Instead of installing a plugin that does fifty things, we write the one thing that matters. A small function. A focused integration. Something that fits neatly into the system instead of expanding it.
Not everywhere—this isn’t about rewriting everything from scratch. There are still plenty of cases where generic software is absolutely the right choice. Payments, email infrastructure, analytics—those are complex for a reason, and you want the robustness that comes with established platforms.
But there’s a whole layer of software that exists mainly because coding used to be expensive.
And that’s where the opportunity is.
A better question
So a better question to ask today is:
“Is this something I should install… or something that’s now simple enough to build?”
It’s a small shift in thinking, but it leads to very different outcomes.
- Fewer dependencies.
- Less overhead.
- More control over how things actually work.
How this shows up with clients
And this is exactly where I tend to step in with clients.
A lot of the work I do isn’t about building massive new systems. It’s about looking at what’s already there and asking:
- Which parts are genuinely valuable?
- Which parts are just historical leftovers?
- And which pieces would be simpler if we rebuilt them—now that the cost has changed?
Often, we don’t remove everything. We just replace the parts where the trade-off no longer makes sense.
The result is usually a system that feels… lighter. Easier to change. Less fragile.
Not the end of plugins or SaaS
So no, AI isn’t killing plugins or SaaS.
But it is quietly taking away the main reason we relied on them for everything.
And once you see that, you start noticing it everywhere.
That extra tool you installed two years ago.
That workaround everyone forgot about.
That “temporary” solution that became permanent.
Some of those are still the right choice.
But some of them? They’re just that drawer full of multi-tools—taking up space, adding friction, and making simple things harder than they need to be.
And now, for the first time in a long while, you actually have a realistic alternative.