The Friction Moved
I told you friction was dead. Here's where it went.
A month ago I told you I built a video game, a tax agent, and a personal assistant. I should tell you which ones I actually still use.
In December I went on a building spree. Five projects in two weeks, fueled by the same thesis I wrote about on this site: the friction that used to stop us from building things is gone. Anyone can turn an idea into something real in an afternoon.
That part was true. What I didn't say is what happens after the afternoon.
The Graveyard
The video game got the most time. 30+ hours across two months. I built the soundtrack first because that was the fun part. The game logic was harder. Then client work picked up, and the game became something I opened every couple of weeks, tinkered with for an hour, and closed. The soundtrack is finished. The game isn't. It's been three months.
The AI assistant was the one I showed people. I had it handling meeting prep, email drafts, client research. I was proud of it. Then someone shipped a better version two weeks later. I switched the same day. Hours of prompt engineering, replaced by a product update.
The slide deck builder was the sneaky one. It worked. Sort of. It could generate 80% of a client presentation. But that last 20%, the part that actually needed judgment, required me to redo most of it anyway. It sits in a repo I haven't opened since January.
Three projects. Three different ways to die. Deprioritized. Obsoleted. Not good enough.
Here's the thing: the video game doesn't bother me. That was for fun. I knew that going in, and I'll finish it when I feel like it. The deck builder is the one that stings. I never decided why I was building it. I just started because I could. Knowing the difference between those two is the skill nobody's teaching.
Every project should start with one of three reasons: to learn something, to have fun, or to solve a real problem. My video game was for fun. I'm fine with 60%. The system my team runs on was to solve a real problem. It shipped in days. The deck builder? I never decided. I just started because I could. That's the failure mode nobody warns you about. Not building the wrong thing. Building without knowing why you're building at all.
The Partial Survivor
The tax agent I bragged about? I still use it. It handles roughly half my tax prep, organizing documents, categorizing expenses, running preliminary calculations. It saves me hours.
I also still use a CPA.
The tool works. I just don't trust it enough to file without a human reviewing everything. And that's a distinction most AI conversations skip entirely. Capability isn't the bottleneck. Trust is. The agent can do the work. I can't verify it well enough to bet my tax return on it.
What Actually Survived
The thing that survived is the system my team runs on every day.
It isn't the flashiest thing I built. It wouldn't make a good demo. It doesn't have a finished soundtrack or an impressive UI. But call prep went from 30 minutes to 5. Deck generation went from 8 hours to 1. Client onboarding went from an hour to 10 minutes. It solves a problem someone on my team was going to have tomorrow morning. And the next morning. And every morning after that.
That's the pattern. The projects that survive aren't the most technically impressive or the most fun to build. They're the ones that started with a problem someone was already complaining about. Not a capability to showcase. A pain to eliminate.
I've started thinking about this as the V0 trap. In my consulting work I talk about V0→V1: ship something incomplete to validate, then refine into something real. But my graveyard projects weren't V0s on the way to V1. They were V0s built for V0 problems, ideas that sounded good but nobody needed tomorrow.
The system that survived started with a V1 problem. Someone needed it today. The V0 version was ugly. It shipped anyway, because the alternative was the pain continuing.
The Mirror
I see the same three failure modes in every organization I work with.
The POC that works beautifully in demo but gets deprioritized because nobody owns it in production. The custom internal tool that gets replaced by a vendor product six months later. The AI solution that performs at 80% accuracy and sits unused because nobody trusts the output enough to act on it.
95% of AI pilots fail to deliver ROI. Not because the technology broke. Because the organization built impressive things instead of essential ones.
My video game and their AI pilot died the same way. The friction wasn't in the building. It was in everything after.
Where Friction Lives Now
The friction isn't gone. It just moved.
It used to sit between you and starting. A team, a budget, six weeks of planning. That friction filtered bad ideas before they cost anyone real time.
Now it sits between starting and finishing something essential. Between building a tool and trusting it. Between a working demo and a daily habit.
The tools keep getting better. Starting keeps getting easier. But finishing, that was always the hard part. We just didn't notice because we rarely got to start.
My original post asked: "What will you build now without friction in the way?"
I had half the question.
What will you use tomorrow?