The AI Productivity Paradox: Why I'm Working More Than Ever
I get more done in a day than I used to in a week. I've also never been this tired.
I had a conversation with a friend last week that I can’t stop thinking about.
Hey, I’m Lakshmi — I help developers build, deploy, and distribute their SaaS without hiring a team. I also run Stacksweller and Supabyoi.
New here? Start with Why Your AI Wakes Up Every Morning With No Memory or Clean Code Is Dead.
We were comparing notes on hitting usage limits with AI coding tools. Both of us on expensive plans. Both of us running into ceilings more often than we did months ago. Both of us, apparently, turning into “power users” in our respective tiers.
And then he dropped this line: “So AI was supposed to make us work less but now we are working more. That’s the conclusion.”
I laughed. Then I stopped laughing.
Because he’s right. I get more done in a single day than I used to accomplish in a week. I’m shipping features, writing content, running experiments at a pace that would’ve been unthinkable about a year ago.
And I have never worked this much in my life.
Here’s what nobody warned us about: AI didn’t give us more time. It gave us more capability.
And capability, it turns out, is extremely addictive.
The Collapse of Activation Energy
Before AI coding assistants, most ideas died a quiet death in my notes app. Not because they were bad ideas. Because the effort-to-value ratio was unfavorable.
“I could build that feature, but it would take a week of focused work. Is it worth a week? Probably not.”
Idea archived. Moving on.
Now that same feature takes a day. Sometimes less. So I build it.
Then I build the next thing. And the next. And suddenly I’m shipping more in a month than I used to ship in a quarter.
The activation energy for starting new work collapsed. And I filled every inch of the newly available space.
Ambition Scales With Output
Here’s the thing about humans: we don’t scope our ambitions in absolute terms. We scope them relative to what feels achievable.
Before AI, I planned projects based on what I could reasonably ship with my limited time and energy. A feature per week. Maybe two if I was focused.
Now “reasonable” means something entirely different. My mental model of what’s achievable expanded by 5x, and my project scope expanded right along with it.
I’m not doing the same work faster. I’m doing more work.
The goalposts moved. And I moved them myself.
The Death of Natural Stopping Points
There used to be friction in development work. Waiting for builds. Context switching costs. The mental load of holding an entire system in your head while debugging.
That friction was annoying. It was also a circuit breaker.
It forced breaks. It created natural pauses where you’d step away, get coffee, maybe realize it was 7pm and you should probably eat dinner.
AI removed the friction. Which sounds great until you realize the friction was also your automatic brake pedal.
Now you can go from idea to implementation to deployment without ever hitting a natural stopping point. The only thing that stops you is your own willpower.
My willpower, for the record, is not great.
The Dopamine Loop of Shipping
Here’s an uncomfortable comparison: AI-assisted coding feels a lot like infinite scroll.
You ship something. It feels good. The tool makes shipping fast and easy. So you ship something else. That also feels good. And there’s always one more thing you could ship.
Same psychological mechanics. Different output.
Except instead of consuming content, you’re producing it. Which feels more virtuous. Which makes it even harder to stop.
“I’m not doomscrolling. I’m being productive.”
Sure you are.
The “Why Not” Threshold
The most insidious change is what happened to my internal cost-benefit calculator.
I used to ask: “Is this worth the effort?”
Now I ask: “Why wouldn’t I just do this?”
That experiment I would’ve skipped because setting it up was tedious? Now I run it. That edge case I would’ve ignored because fixing it properly would take half a day? Now I fix it.
The threshold for “worth my time” dropped to near zero. So everything is worth my time. So I do everything.
This is how you end up working 12-hour days while technically being more “efficient” than ever before.
The Uncomfortable Truth
AI tools didn’t give us more free time. They gave us more output capacity. And we’re psychologically incapable of leaving capacity unused. At least I am.
The work expanded to fill the available capability. Parkinson’s Law, but in reverse.
We’re not working less. We’re shipping more while feeling productive. Which is a different thing entirely.
My friend was right to put “off” in scare quotes when wishing me a good weekend. We both knew I wasn’t really taking time off. I was just switching to a different kind of work.
What Now?
I don’t have a tidy solution here. I’m not going to pretend I’ve figured out work-life balance in the age of AI assistants.
But I’ve started noticing when I’m filling capacity just because I can. When I’m starting a new feature not because it matters, but because the activation energy is so low that “why not” won the argument.
Sometimes the answer to “why not” is: because you could just... not.
Groundbreaking insight, I realize.
The AI isn’t going to set boundaries for you. If anything, hitting usage limits might be the only forced break some of us get. Which is both sad and a little funny.
Maybe the real productivity hack is learning to leave capability on the table.
I’ll let you know how that goes. Right after I ship this one more thing.
I write about building and deploying software as a solo developer. If you’re trying to do it all yourself without hiring a team, I’m probably making the same mistakes you are.