Last week the topic on everyone's mind seemed to be the future of programming. All my friends were talking about it, people I follow were talking about it, and I was thinking about it as well. Then David's (DHH's) Rails World keynote turbocharged the conversation.
His talk didn't come as a surprise to me. I was a week into rewriting HEY in Rust during Rails World. And "Pencils down" only codified how we'd been working for months at that point.
David is right. Like it or not, we programmers, like painters before us, are slowly going to disappear. AI, like the camera, completely changed the game. You can sulk about it, or you can pick up a camera, start learning, and become a photographer.
The perfect blend
I was skeptical of AI and vibe coding. My initial dabbling with it didn't turn out that well. The agents needed a lot of hand-holding to produce a decent result. But the models improved at an incredible pace, and I got better at employing them, learned their limitations, and learned how to express what I wanted.
For me, the first real test for this new way of working was HEY CLI.
Rob and I built it - the CLI, TUI, and the supporting API, SDK, and ActionCable client - over the span of a week while he was traveling, and while I was shipping and polishing a major pricing change for Basecamp. It's completely vibe-coded. No human has seen any of the machinations in the code beyond the initial project setup - and it worked out!
Something clicked for me on that project. I suddenly realized that my only limit now was my imagination.
Jason joined us late in the week to turn our programmer art into something delightful. He did wonders in just a day - in a completely foreign medium (ASCII art), in a programming language he doesn't know, in a codebase no human has ever read.
That got me thinking: "Maybe I could design the calendar?"
I had a vision for what the finished calendar TUI could look like, so I gave it a go.
Designing the calendar was challenging. There were dead ends and abandoned ideas, but I got it done, I'm very happy with it, and I leveled up as a designer.
My programming skills helped me along the way. I caught things that wouldn't work and guided the agents towards better solutions. I didn't need to write the code myself, but knowing how software works still made me better at getting the result I wanted. And the same was true for design. The more I learned about typography, layout, and terminal interfaces, the more things I could imagine trying.
AI removed the speed limit, but it didn't remove the limits of my imagination. The more you know, the more you can imagine.
Walls between previously strictly defined roles have broken down: programmer, ops, designer, QA, mobile, web. This doesn't mean expertise is useless. Quite the opposite. A lot of what I learned over the years carries over and helps me see possibilities, problems, and tradeoffs I otherwise wouldn't. Merlin Rebovic gave a great talk about that.
But knowing a bit of everything gets you much further today than being excellent at one or two things. Every new skill gives you another way of thinking, another set of things you know to look for, and, most importantly, more things you can imagine. The future needs a more even blend of skills.
Eventual excellence
What if I told you that the photo above is the third shot out of five? Would that take away from the photo? Is it somehow tarnished by not being perfect on the first try? By the two photos before and after it?
Of course it isn't. Nobody looks at a photograph and asks: "Yes, but how many bad exposures did you take?"
Misfires, variations, and riffing are part of the process. It takes taste to pick out the best one and to polish it up.
It's the same when working with agents.
We used to turn an idea into code, then meticulously polish that code to excellence, and ship it. Every idea had to prove its place; every line of code had to be worth the time it took to write and think about it. Writing code used to be expensive, but it isn't anymore!
Now I can turn an idea into something real in minutes, play around with it, get a feel for it, and then iterate on it or throw it away. The first attempt will probably be rubbish. That's okay. Every attempt and iteration teaches me something about where I want to go next.
There will be false starts, rabbit holes, and bugs. There will be things that fall through the cracks. Excellence doesn't have to happen on the first try. It can be eventual.
I can try five ideas in the time it previously took me to prove just one. Four might not work out. But, just like the four photos you never saw, that doesn't diminish the one that worked.
The goal is still excellence. We just don't need it at every step of the way anymore.
Do civil engineers do engineering?
If I'm no longer writing the code, what exactly am I doing? If you ponder that question, let me ask you another one: Do civil engineers do engineering?
My mom is a civil engineer. I have tons of memories of her at her giant drawing table with rice paper, rulers, pens, and razor blades, designing people's homes. With her pen she dug trenches, laid rebar, set walls, and picked the type of concrete to use without ever picking up a shovel. She was doing engineering. She was solving hard problems with hard constraints. The fact that someone else picked up a shovel, cut the rebar, set up molds, and poured the concrete doesn't take that away from her.
So don't conflate writing code with engineering.
With the agents pouring concrete, we can now focus on the thing as a whole. Not just how to implement it, but how to design it, test it, deploy it, make it accessible, bring it to other platforms, announce it, promote it, and charge for it. Instead of designing walls, we'll design homes.
Programmers will disappear, but engineering software won't. It will morph into something broader, something that's a little bit of everything. We have to become makers.
Closing thoughts
Picking up a camera won't instantly make you a photographer. Most of us walk around with a great camera in our pockets, yet few of us are actually good at using it.
If you take a single picture, compare it to the subject, and conclude "this is rubbish," that doesn't necessarily mean the camera is bad. You have to learn how to use it. Learn its limitations and how to work around them.
Photography is also a skill. It has to be practiced, explored, and honed.
Like painters in the 19th century, we are in uncharted territory. Nobody really knows what they're doing or what being good at photography even means. We're all exploring this new space and trying to figure things out, some with more success than others.
There's only one way to find out where this goes: pick up the camera.