Calling the Plumbers

I have a confession. I like to look at myself in the mirror more than I like to worry about the plumbing.

Not literally. But when it comes to building things, I’m drawn to the parts I can see. The surface. The feel. The way a product presents itself to the world. Give me brand positioning, marketing copy, interface polish. I’ll go deep on those for hours.

The backend? The data pipelines? The authentication flows? Those exist in a place I can build but don’t love living in.

The Webflow years#

No-code gave me something specific: permission to build without heavy scrutiny. I could play. I could learn. I could ship things that looked and felt professional while figuring out how they actually worked. Nobody was watching my code because there wasn’t code to watch.

That environment taught me to care about craft on my own terms. My integrity wanted me to be better at it, not somebody else’s code review. I pushed myself because I knew there were better ways to do things, not because someone was looking over my shoulder.

It’s a different kind of accountability. Softer in some ways, harder in others.

Enter the AI era#

When Opus landed, I thought the calculus might change. All those backend things that felt tedious or out of reach suddenly seemed possible. And they are possible. I can build them now.

But here’s the thing I didn’t anticipate: possible isn’t the same as motivating.

The architectural work, the API integrations, the testing loops that surface success through logs instead of visuals, all of that still doesn’t drive me the way a good interface does. AI didn’t change what I find interesting. It just expanded what I could technically accomplish.

I spent the last few months trying to be two different kinds of developer. The one who polishes the surface and the one who runs the pipes. Turns out I still only want to wear one pair of shoes.

Shiny objects and hound dogs#

I have a pattern. An idea grabs me and I’m gone. Full hound dog mode.

I’ll spend weeks dialing in the brand, the tone, the audience, the product feel. I’ll explore features through the lens of marketing, intuiting what people need and how they’d need it to work. I’ll validate positioning before a line of backend code exists.

This is actually a skill. I’ve started thinking of it that way instead of as a problem to fix.

The problem isn’t the exploration. It’s what happens after. Eventually the shiny part is done and what’s left is the plumbing. And I either force myself through it, taking forever and burning out, or I jump to the next shiny thing and leave infrastructure half-built.

I’ve tried time-boxing. I’ll work on this for two weeks, then move on. I set limits. I don’t follow them. When the idea has me, I’m terrible at saying no.

I haven’t figured out the right harness for that energy yet. But I’m starting to think the answer isn’t better discipline. It’s better partnerships.

There’s a reason we have plumbers#

A friend said something that stuck with me: most knowledge experts are more than happy to help when you ask.

I’ve been a solo practitioner for most of my career. Not by ideology, just by circumstance. Freelance work is often one person doing everything, and no-code made that sustainable longer than it might have been otherwise.

But the AI shift is highlighting something I could ignore before. The work I’m avoiding isn’t going away. It’s getting more important as what’s possible expands. And forcing myself to care about it isn’t working.

The move might be to build things to the edge of my abilities, then admit where I’m stuck and ask for help. Not delegate everything. Not hand off a napkin sketch. But get something real enough that an expert can look at it and say “here’s what I’d change” instead of “here’s where I’d start.”

That feedback loop might be enough. Take their input, throw it at a planning session with an agent, iterate. I don’t have to love the plumbing. I just have to not be precious about owning all of it.

Integrity and shipping#

Here’s the tension: I can’t ship sloppy work. Even when the sloppiness is invisible to users, I know it’s there. Security, performance, decisions that will bite future-me. The craft thing that made no-code work for me is the same thing that makes half-finished backends feel wrong.

Some people can throw up an MVP with duct tape infrastructure and iterate from there. I envy that. My version of “rough draft” is more polished than it needs to be, because polished is how I think.

So the collaboration can’t be “just ship it, we’ll fix it later.” It has to be “let’s get this to a standard I can live with, together.” Different shoes, same level of care.

What I’m actually good at#

I’m sitting with this framing: maybe my job isn’t to become a full-stack builder. Maybe it’s to be really good at the front half.

Product feel. Brand positioning. Feature intuition through the marketing lens. Finding what people need before they know they need it and figuring out how it should feel to use.

Those skills still need implementation. But they might be more valuable as inputs to a team than as the first half of a solo operation I never finish.

I’m not sure what that looks like yet. I’m still figuring out how to put myself out there in a way that surfaces these perspectives and does some of the legwork on finding the right collaborators.

But I’m starting to think the answer to “how do I build everything myself” might be “you don’t, and that’s fine.”

Finding footing#

Everyone I talk to is navigating some version of this. The tools are changing fast. What used to require a team can sometimes be done solo. What used to be impossible is now just tedious. The lines are blurry.

We’re all trying to figure out where we fit. What to hold onto, what to let go of, what to learn and what to hire out. Some days that feels exciting. Other days it just feels like a lot of adapting to do.

I’m giving myself more grace about not having it figured out. And I’m trying to let go of the ego that says I should be able to do it all myself.

The plumbers know things I don’t. It’s probably time to make some calls.