Building got instant. Everything after it didn't

I’m a person who really likes instant gratification. I’d rather say that up front, because it explains most of what follows.

For years the slow part of my side projects was building them. Now building is the fast part, and I’m finding out what was hiding behind it.

Thanks for reading! Subscribe for free to receive new posts and support my work.

When building was the bottleneck

For six or seven years I ran Qesheq, an Arabic writing platform, something like Medium. I built it by hand, beside a full-time job. Every feature took weeks of evenings. The editor and the writer accounts took so long that by the time I had a decent version, I’d lost hope. I took the site down a year or two before Cursor existed.

Back then I had the perfect excuse for anything that didn’t work: I didn’t have time. If I had the time, I’d build the missing piece, and then it would work. There was always a feature to blame.

What last week looked like

Today I build Relnk, a writing studio, and Naseeha, beside the same kind of day job. Since the end of December I haven’t written code by hand. I have a command that takes a ticket, builds it, reviews its own pull request, fixes the comments and puts it on staging. What’s left for me is QA, the deploy, and deciding what’s worth building.

Last week I wanted a better sign-up page for Relnk. Instead of building one and guessing, I had two built side by side: the old page, and a new one that leads with the trial, puts Google sign-in first, asks for less, and says plainly what the AI does. Both send the same events, so I can follow people through either one. I clicked through both on staging, picked the new one, and the old one was gone two days after the first commit.

When I built by hand, that would have been one page and a guess. Now it was a decision I made in front of the TV.

Most of the work doesn’t even need my laptop anymore. Claude Code runs all the time on a small server, so I can start a ticket from anywhere and check the result later. The building happens whether I’m at my desk or not. My job is to show up with a clear idea of what should exist, and to look properly at what comes back.

The part that didn’t speed up

Here’s what didn’t change: people still have to find the page, read it, and decide. That runs on their time, not mine.

I can ship a feature on Tuesday. I can’t make anyone trust a product by Wednesday. Conversations, replies, the first people who try it and tell you what’s wrong: none of that got faster because Claude Code got better. If anything it feels slower, because everything around it is so fast now.

And that’s exactly where I’m weak. The thing most likely to make me drop a project was never time. It’s not seeing results. One, two, three, four weeks of nothing, and I start losing the passion.

Slow building used to hide that from me. There was always one more feature standing between me and the moment I’d find out if anyone cared. Now I get to that moment in days, and then I’m just waiting. Refreshing a dashboard is not a plan, but it’s very tempting when the alternative is sitting with the fact that the next step isn’t mine to speed up.

What I’m doing about it

Honestly, I’m still working it out. Three things so far.

I stopped counting shipping as progress. Shipping is cheap now, so it tells me very little. A week with no new features and one real conversation with someone who’d use the thing is a better week.

I reply more than I post. Most of my time online right now goes into replying to other people where they already are, not into my own posts. It’s slow, and nobody notices for a while. It’s also the only thing I’ve found that builds anything with people who don’t know me yet.

I wrote the quitting rules before launch. Relnk’s launch plan has kill rules: for each channel and each stage, what result counts and how long it gets before I drop it or change it. I wrote them while I was calm. Now a quiet Tuesday night doesn’t get to decide for me. The date does.

None of this makes the waiting feel good. It just stops me from filling it with features nobody asked for, which is what I’d do by default. When building is this easy, building is the most comfortable way to avoid finding out.

What would change my mind

If the slow part speeds up too, I’m wrong. If agents start finding the right people and earning their trust as fast as they build, then the gap I’m describing closes and the old game comes back.

I’d bet against it. Building got instant. Trust still takes as long as it takes.

Thanks for reading! Subscribe for free to receive new posts and support my work.