Claude has gamified work

So I barely leave the computer

It’s 11:36PM (two hours past my bedtime) and I’m staring at the screen bleary-eyed, bashing the enter key to let Claude Code execute another task that will only take 30 seconds.

Except I said this to myself at 9PM, and again at 10, and again at 11.

Work has become so easy, so effortless, so vibey. Four lines of articulate instructions could result in a whole new webpage. A fleeting idea could become a web app. An urge to reorganise my second brain could actually reorganise my second brain.

All this done through the Terminal, which was once the most boring application on my laptop.

I started with one Terminal window. I talked to Claude in natural language (aka English), then carefully observed what it was doing to the files in my finder and to the blocks in my Notion page. It didn’t take long to get bored while waiting for it to sauté and frolic around. So I opened another Terminal window, then another one, before the whole screen was covered in Terminal windows.

They require my input, so it feels like playing whack-a-mole with the different tasks. Downloads folder organised? Great, it can move on to screenshots next. Research for a YouTube video idea done? Great, it can now suggest 10 titles. Small Creator Big World newsletter recap drafted? Time to generate an image then.

This very quickly became my life, an all-consuming series of strings on a slate screen.

But boy was this fun. Claude drops me into a state of flow. Every single time.

In Super Mario, you start off at level 1-1, going at a slow pace, jumping up to break the brick boxes and collecting mushrooms to power up. Then you get to level 1-2, and you start having to jump on turtles to defeat them. You become better at the game, so the game increases in difficulty and you feel constantly challenged. You keep going, because the dopa hits when you defeat a boss level right at the edge of your capabilities. And so the game goes.

Working with Claude feels like that. Claude learns how you work, so it can execute tasks faster. You learn how Claude works, so you ask it to do harder things. You start with one web app, then three, then seven. Then you set up an Openclaw agent. Then you fiddle around with automations. And so the game goes.

When I first tried AI, I was enticed by the promise of it speeding up the work too repetitive, too boring, not challenging enough. Now that I’m there, I couldn’t help but cram even more work into my 8-hour workdays. And then I get excited about everything I can do, so I tinker with Claude in my evenings, then in my weekends.

I don’t know what to do with this. On one hand, I’ve never found work more fun. Claude blows past my technical limits, which unlocks decision-making, speed, and shipping. But on the flip side, I pretty much only work.

Even so, I have my human limitations (aka how many hours I can stay up). So I tell Claude to get on with a more complex task, just long enough for me to switch off my study lights and let exhaustion take me to bed. Tomorrow, I’ll review the changes and pick up where I left off.

And so the game goes.

---

Update log:

• 📖 Started reading Empire of AI by Karen Hao per a recommendation from a friend

• 🎙️ Batch recording a bunch of Small Creator Big World episodes ahead of a hectic travel season

• 📙 Finished The Emperor of Gladness by Ocean Vuong and OOF. The ending hit. Or rather didn’t hit. And so it’s lingering

• 🌈 I’ve been working on getting prism.lgbt live - it’s a directory for LGBTQ+ spots in HK. Finally had time to work on this properly and I’m so glad this is now out in the world

• 👋 Caught up with a wonderful long-time friend who sent me a pic of my art book with her baby 😭

• 🤖 Trying out this agent-in-Slack (kinda like Openclaw but it actually works) called Viktor and it’s really good

Work with me: https://go.beckyisj.com/workwithme

Some links are affiliate links, meaning that I may receive a commission if you make a purchase through the links at no cost to you.