My Abstraction Crisis: Staying Sane While AI Keeps Moving the Goalposts
This is the new tax on builders: you are not just building products, you are rebuilding the abstraction you build with.
I have never been able to build this fast in my life, and I have never felt this unstable.
The hardest part of building with AI right now is not writing code. It is watching something you were excited to build become “already handled” shortly after you start. Not because you failed, but because the level of the game changed once again.
That mental whiplash is what feeds almost daily into my abstraction crisis. AI has changed the economics of building: execution is cheaper, tooling is denser, and what used to take weeks now takes hours.
So my bottleneck is not output anymore. My bottleneck is direction. What is worth building when agents can turn half my backlog into working drafts overnight? What do I bet on when the foundations keep shifting in real time?
The scary part is not speed. It is what speed does to my attention, my confidence, and my life.
This is the new tax on builders: you are not just building products, you are rebuilding the abstraction you build with.
The moment it hits
One day I’m wiring tools together and I feel ahead. My workflow is smooth, my pace is high, and it feels like I found compounding leverage.
Then the world shifts. A release drops. A capability lands. An integration becomes default. Suddenly what felt like an edge yesterday feels ordinary today, and I wake up with that tight, anxious feeling of being behind.
Here’s a concrete example. One of my recent side quests is building a “brain funnel,” a system to capture, sort, and resurface thoughts and ideas. Then I stumble across David Szabo-Stuban’s agentic butler and realize there’s already a high-fidelity repo of the thing I’m building, basically sitting there, finished enough to make my version feel unnecessary.
It creates a cycle that is hard to admit. One week I feel unstoppable and I am stacking progress across projects. The next week I feel small, not because I stopped working, but because the baseline moved again.
This is what the abstraction crisis does to me. It makes commitment feel naive. It makes depth feel risky. It tempts me into living at the surface, constantly scanning for what changed. And that is exactly when I lose the thread.
Focus compounds
Here is the part I keep relearning the hard way: when the tools speed up, my brain tries to speed up with them. I want to update everything, rebuild my workflow, and start five new projects because the frontier moved and I can “obviously” do more now.
That instinct feels rational. It could also be a trap. The thing that actually compounds is not novelty, not constant reorientation, and not being first to duct tape the newest tool into a workflow.
The thing that compounds is focus.
Focus turns a week of excitement into a year of progress. It lets me build something deeper than a demo, something with taste, structure, and a point of view. It keeps me from rebuilding my life every time the tooling shifts.
Some people experience focus as limitation. For me, it is essential for survival.
If AI makes execution cheap, then attention becomes the scarce resource.
The real challenge is staying oriented
Every day it feels like I am being pushed up a level. Tools become workflows, workflows become systems, and systems become meta systems (systems that manage other systems). Then the baseline shifts and it happens again.
The temptation is to treat this like a race. Keep climbing, keep upgrading, keep stacking leverage.
But the real challenge is not output. It is whether I can keep climbing abstraction layers without losing my life to the climb.
My definition of winning is not “most shipped.” It is a good life: balanced, connected, healthy. Books. Runs. Games with family and friends. Laughing and crying, together.
So I needed constraints, not motivation.
Two modes, one abstraction budget
What helped me more than any productivity hack is splitting my building life into two modes, and giving myself an abstraction budget.
Frontier mode is where I explore: new models, new agentic workflows, new ways of wiring things together. This is where I play at the edge and learn what is possible.
Production mode is where I execute: I ship in a chosen lane with as little novelty as possible. No rebuild because a new feature dropped. Just making something real.
The abstraction budget is what makes those modes work. I have a limited capacity for climbing abstraction layers without frying my attention and leaking into comparison. When I spend it, I feel it. I start second guessing. I start scanning instead of building. I start living in reaction mode.
So I treat frontier as spending the budget, and production as protecting it. Practically, that means I cap frontier time, I do not let frontier invade production because of hype or fear, and I do not confuse “I tried the new thing” with “I made progress.”
So I try and stick with projects that feed my soul. Projects that teach me depth I actually want. Even if other people shipped something similar. Similar is not the same. My taste is not their taste.
“But if I don’t maximize speed, I’ll lose.”
I get it. The pace is real. Leverage compounds. Windows close.
But speed isn’t the same as progress. Velocity without direction is expensive chaos.
The point isn’t to slow down. It’s to aim.
Staying oriented
AI will keep moving the goalposts. The ladder will keep extending. Agentic workflows will keep getting absorbed upstream.
I can respond by chasing every new abstraction the moment it appears. Or I can anchor to what compounds even when the tools change: focus, depth, and a life I actually want to live.
When I feel the churn starting, I go for a run before I change anything.
And if I am honest, that is what this is really about. Not productivity. Not tools. But protecting the parts of my life that cannot be automated, so the life I am building actually belongs to me.