Add "Endless Work" blog post (#1)

New post about how AI collapses the activation energy for starting tasks,
creating an endless cycle of planning and shipping at startups.

Co-authored-by: Cursor <cursoragent@cursor.com>

Author: berniexie

src/blog/posts/endless-work.md

@@ -0,0 +1,39 @@
+---
+title: "Endless Work"
+slug: "endless-work"
+author: "Bernard"
+date: "2026-02-18"
+description: "AI has collapsed the activation energy for starting new tasks, turning startup engineering into an endless cycle of planning and shipping with no natural break points."
+tags: ["ai", "startups", "burnout", "productivity", "health"]
+draft: false
+---
+
+Here's a strange phenomenon I've noticed as I've gotten more proficient with LLMs: I can plan my next task while I'm still building the current one. That sounds like pure upside. It is not.
+
+What actually happens is this. I'm mid-feature, I hit a bug, and while the agent is fixing it, I start scoping the next thing. Then I notice something else that's broken, so I kick off a plan for that too. By the time I ship the original feature, I'm already deep into two or three other workstreams. There's no natural stopping point. No seam in the work where I can step away. I just keep going, pulled forward by the momentum of whatever I already started thinking about. It honestly feels like an addiction. You're hooked on the high of pushing the next change.
+
+## This isn't new, but it's worse now
+
+At a startup, the assumption is that everything is always broken. Customers want new features. There's a backlog. And then there's the stuff that surfaces organically while you're building: edge cases, performance issues, "oh wait, this other thing is wrong too." In the past as a founding engineer, I was already drowning before AI entered the picture.
+
+But the activation energy for starting something used to be much higher. You'd finish a big PR or squash a brutal bug, and there was a moment of natural rest. Not a vacation. I'm talking about getting a glass of water. Walking outside. Doing literally anything that isn't staring at a screen. Those moments existed because starting the next thing required real effort: reading docs, sketching an approach, context-switching into a new part of the codebase.
+
+AI removed that friction almost entirely. I can plan and execute at the same time now, and those small recovery windows have vanished. I find myself hunched in a chair for 12, 13, 15 hours a day, continuously shipping code. That's great for my company's velocity compared to where I was even a year ago. It is terrible for me.
+
+I need to figure out how not to burn out. Burnout is weirdly glorified among startup founders (I remember batchmates at YC bragging about coding from the hospital), but building a company is a marathon. If you flame out in the first few miles, you never finish the race.
+
+---
+
+## What I'm trying
+
+These aren't groundbreaking. They're just important.
+
+1. **Always take walks.** Go outside at least once a day. My personal rule is to hit 10,000 steps no matter what. If it's 11:30 PM and I'm short, I'm going outside.
+
+2. **Protect your sleep.** Founders work late, and sometimes that's unavoidable. But sleep deprivation compounds. The code you write at 2 AM on four hours of rest is not good code. It's like studying: past a certain point, you're just burning hours for diminishing returns.
+
+3. **Keep a non-engineering hobby alive.** Work the parts of your brain that think differently. For me that's lifting, drumming, tennis, cooking. Beyond the cognitive benefits, so much of leading a good life is being a multidimensional person. If all you know and talk about is your startup, you won't be interesting. You'll just be annoying.
+
+---
+
+I'm curious what other people are doing to break this cycle. It feels genuinely new, a kind of work pattern that could not have existed before AI collapsed the cost of starting the next thing. If you've found something that works, I'd love to hear it.