← Back to list

AI Made Me 3x Faster. My Eyes Sent Me the Bill.

I shipped more in six months than in the two years before it. Then I noticed I couldn’t read a street sign at the end of my block. Here’s…

Raphael Ben Hamo · 2026-08-23 15:46 · 0 claps · 5.3 min read
#eye-strain #ai #technology #tech
Open on Medium ↗
Wiki topics: AI · AI · General

AI Made Me 3x Faster. My Eyes Sent Me the Bill.

I shipped more in six months than in the two years before it. Then I noticed I couldn’t read a street sign at the end of my block. Here’s what I got wrong — and the embarrassing part: I build an eye-care app.Second Heading

Quick disclaimer before anything else: I’m a developer and an indie founder, not a doctor. Everything below is my own experience, my own tracking, and my reading of public research. It is not medical advice. If your eyes hurt persistently, go see an optometrist — I did.

The good part

Sometime around late 2024 my job quietly changed shape.

I stopped typing most of my code. I started reading it. Diffs, plans, generated specs, agent output, five-hundred-line refactors I had to review in one sitting because approving them blind is how you ship disasters. My commit count went up. My shipped-feature count went way up. On paper I became a three-person team.

What nobody put on the slide deck is what that does to the physical act of working.

Writing code is slow and lumpy. You type, you stop, you stare at the ceiling, you get coffee, you argue with yourself. There’s dead air in it, and dead air is where your eyes rest.

Reviewing AI output has no dead air. It’s dense, continuous, high-stakes reading at 60 centimeters, for hours. The thinking pauses are gone — the machine already did the thinking, and now you’re the bottleneck, and you can feel that you’re the bottleneck, so you don’t stop.

I didn’t work more hours. I worked the same hours with all the gaps removed.

The bill

It showed up in a stupid way.

I walked out to grab lunch and the sign on the corner café was soft. Not blurry-blurry. Just… unresolved, for a couple of seconds, and then it snapped in. It happened again that week. And again.

That specific thing has a name — your eye getting stuck in near-focus and taking a beat to release. Inside your eye there’s a small ring of muscle, the ciliary muscle, that squeezes to focus on close things and relaxes to focus on far things. Screens keep it squeezed. Hours of unbroken near work keep it squeezed for hours. Reviews of digital eye strain put accommodative fatigue — that muscle running out of road — near the center of why screen work hurts.

There’s a detail in that literature I can’t stop thinking about: the eye’s focusing effort tracks cognitive load, not just distance. Harder mental task, more accommodative activity. Which means reviewing a gnarly AI-generated migration isn’t merely “screen time.” It’s screen time with the difficulty dial turned up.

So: I tripled my reading, removed my pauses, and raised the difficulty. Three multipliers, same pair of eyes.

Meanwhile the surveys were catching the systemic version of this. 83% of knowledge workers reported burnout in 2026, and — the part that should stop you — Harvard Business Review data has 88% of heavy AI users reporting increased burnout. Heavier AI use, more burnout, not less. The tools gave back time and we immediately spent all of it on more tool-mediated work.

Everyone’s optimizing the wrong layer

Look at what the 2026 productivity conversation is actually about: context windows, model routing, prompt libraries, agent orchestration, which IDE, which harness. All throughput on the machine side.

Almost none of it is about the throughput of the human doing the reviewing. And when we do talk about the human, we jump straight to the abstract layer — focus, motivation, “protecting deep work.”

The physical layer is sitting right there, unmeasured. You have one input device for all of this and it is a pair of small wet organs with a muscle in them that fatigues like any other muscle. You wouldn’t triple your deadlift volume overnight and then wonder why your back hurts. We did exactly that to our eyes in about eighteen months and called it a productivity revolution.

What actually moved the needle for me

I want to be careful here, because this is where posts like this usually go off a cliff into supplements and gadgets. Three things helped. None of them cost money.

1. Breaks tied to the clock, not to my sense of when I need one.

This is the whole thing, honestly. The 20–20–20 rule — every 20 minutes, look at something ~20 feet away for ~20 seconds — is not exotic advice. Every optometrist says it. The reason it doesn’t work isn’t that it’s wrong. It’s that the moment you most need it is the moment you’re least able to notice you need it. Deep in a review, twenty minutes becomes ninety. Flow beats willpower every single time, and honestly, good — flow is the point.

So it has to come from outside your head.

2. Distance, specifically. Not “a break.”

Standing up and checking Slack on your phone is not an eye break. Same focal distance, same muscle, same contraction. The variable that matters is distance, which means a window, a hallway, the far wall — something across the room or across the street. Twenty seconds of actual far-focus does more than five minutes of walking around still looking at things 40cm from your face.

3. Blinking on purpose during long reads.

Blink rate drops hard during concentrated screen work — roughly 15 a minute down to about 5. Nobody can consciously manage this all day, but I can do it for the twenty seconds I’m already looking out the window, and that’s most of the benefit for none of the effort.

The embarrassing part

Here’s the thing I’ve been circling.

I already build an app for this. It’s called EyeReset. It runs a 20–20–20 timer, it has guided eye exercises, it tracks whether you actually took your breaks. I shipped it, I wrote the App Store copy, I know the research well enough to argue about it.

And I was, without question, the worst user of my own product. It was installed on my phone the entire time my eyes were falling apart. I had built the tool and skipped the habit — because I was busy, because the sprint mattered, because I’ll do it after this one review.

That’s not a cute humblebrag. It’s the actual lesson, and it’s why I think most eye-strain advice fails: knowing the rule contributes almost nothing. You already knew about 20–20–20 before you opened this article. Knowledge was never the missing piece. The missing piece is that the reminder has to arrive from outside your attention, at a moment you have already decided in advance, when the version of you that cares about your eyes is not the version currently at the keyboard.

Once I actually turned the reminders on and let them interrupt me — grudgingly, resenting them for about four days — the café sign stopped going soft. The 4 p.m. fog got noticeably thinner. I’m not going to claim more than that, because that’s all I measured.

If you take one thing

Your AI stack made you faster at the thinking. It did nothing for the hardware doing the reading. That gap is where the 2026 burnout numbers are hiding, and it’s a gap you can close in twenty seconds at a time.

Pick a distance you can look at from where you sit right now — a window, a doorway, the end of the room. Then make something other than your own memory responsible for sending you there every twenty minutes. A timer, an alarm, a sticky note, a co-worker. Or the app I built precisely because I knew I’d never remember on my own.

👉 EyeReset is free on the App Store — 20–20–20 timer, guided exercises,


메타데이터
post_id
c1eb1031c404
slug
ai-made-me-3x-faster-my-eyes-sent-me-the-bill-c1eb1031c404
url
https://medium.com/@rafisem7/ai-made-me-3x-faster-my-eyes-sent-me-the-bill-c1eb1031c404
canonical_url
https://medium.com/@rafisem7/ai-made-me-3x-faster-my-eyes-sent-me-the-bill-c1eb1031c404
author_url
https://medium.com/@rafisem7
status
ok
fetched_at
2026-08-25 11:47:52