I found myself revisiting a video game from my youth recently: Final Fantasy X. If you haven't watched the intro cutscene, do yourself a favor and take three minutes to watch it and come back when you're done.

The game is a literal hero's journey. It sucks you into the story right away and shows the hero being pulled into the threshold of a new world by his mentor. This is happening whether he likes it or not. The mentors emphasizes that, “This is it. This is your story. It all begins here”, before they are pulled into another world.

I've always heard that we're largely driven (one could also say controlled!) by the stories that we have of ourselves and our heads. Many of these stories are told to us before we can even identify the component parts of the story, but they are formative regardless. As I get older, I see that more and more clearly. The perspective we take and the characters we identify with in these different stories change over time, too.

When I first watched the above FFX intro cutscene when I was 11 or 12, I largely identified with the protagonist who saw this otherworldly [1] event as something that's happening to him that he didn't get to choose. Yet, we spend the rest of the game literally playing as him, navigating this new world, building autonomy through completing quests, fighting monsters, making friends, and eventually saving the world.

It is a very typical “reluctant hero” character arc that you've seen in any number of movies like Star Wars, Lord of the Rings, etc.

Now, I'm starting to feel more and more like the Unc that is Auron vs. the youthful protagonist of the series, Tidus. For context, Auron is an older, grittier, “been there, done that” type of character. He overlooks the city, staring down a looming reality-shattering threat while the youth are preoccupied with their sports ball. Sound familiar?

I wanted to have an excuse to talk about Final Fantasy X in a professional context, but also it fits because we are getting taken to this other world where coding with AI agents is the norm, whether we like it or not. It is up to us how we decide to react to this new world. We have control over the story of how AI is affecting / will affect us in our own heads.

From what I've seen in my own usage of AI Agents, it enables disproportionately more discretionary effort. Meaning, I choose to work harder for longer on something to get better results. This rhymes with something that I've also seen recently.

I’ve been following Predictive History for a few months, almost “before it was cool”. Many others recently have also been put on to this professor guy about his predictions on the Iran war. Whether he is a dressed up conspiracy theorist, CIA psyop, skull and bonesman, etc or not, he makes an interesting point about the ultimate power to wield in this world is human consciousness.

Everything relating to politics (which is everything) revolves around harnessing the human spirit, energy, attention, etc. Nation states are just lines drawn around land, resources, and groups of people to harness the collective power of the combination thereof. Social media sites are more or less the same thing, just more narrowly defined with attention.

I believe AI Agents are a new way to harness human attention in a way that leads them to be much more productive than without. Agents may be a way to cope with brainrot or may just be a way to speed run further brain rot faster. Either way, I'm able to parallelize and juggle more work more than I ever have before, while also being more engaged and having more fun than I ever have at any point in my career.

When I have fun doing something, I want to spend more time doing it, and the better I get at it, which leads to more fun. This is a virtuous cycle. I experienced this first with leveling up in video games (like FFX) but in a deeper and more profound way with sports, most notably Brazilian Jiu-Jitsu, which I have trained in for the past 13 years.

Get out your bingo cards, here's the Jiu-Jitsu analogy for this newsletter. You level up in Jiu-Jitsu based on belts (and stripes as mini-levels). That's why you start doing it. You want to get a black belt or some other external validation, but then somewhere along the way you fall in love with the process (or you don't, and you quit), and it just becomes a part of what you consistently do. Who you are is what you consistently do.

Even if there were no belts, I would continue to do Jiu-Jitsu regardless because it is fun and I can feel myself improving in it over time. I feel a sense of mastery of fundamental truths such that I can teach people how to get better at it over time as well.

AI Agents are automating away the crappy parts of figuring out syntax and looking up documentation, but are they also going to automate our entire jobs away? I don't think so.

I've had a hunch that consistently using the same agent over time is sort of like getting used to a very particular type of tool, such as you have your favorite hammer, golf clubs, baseball mitt, or even your daily driver car. You just get used to the feeling of the tool over time and it just melts into your perception. The tool becomes an extension of you. This is well-documented science.

I'm starting to feel the effects of this, given a long enough time frame with my agent of choice, Claude Code. When faced with potentially running out of usage, I get a little bit of anxiety because I know other agents aren't going to react in the same way that I know how Claude will react.

But at the same time, I don't feel like I need any other tooling or frameworks or even a real AGENTS.md file to guide Claude. I'm very comfortable with Claude Code’s native built-in tooling, like planning + tasks, and knowing which models to use when.

/model opusplan is a great way to get the better analytical capabilities of Opus while executing and implementing with the cost-saving Sonnet model. At the same time, if I run into a hiccup or an issue, I need to go back to Opus to get to the bottom of it with debugging.

I'm not a big skills person, and maybe I don’t have to be? Or maybe I’m missing out on a big upgrade, idk. I feel pretty productive without them, but maybe I need to look to continually improve.

I've only learned how to use all this stuff in the Claude ecosystem because I've stuck with it. If I cobbled together a couple of $20/month usage tools, I would not be as comfortable nor productive with one, but maybe I could see parallels between multiple.

I'm not going to argue with Bruce Lee. If I can continue to develop my skill at wielding one of the most state-of-the-art problem solving tools, which the rank changes weekly, I'll be okay moving forward.

I think whether or not we get automated out of jobs comes down to whether you see the number of problems in the world in general as decreasing, flat, or increasing. Only one of these is supported by a 90s gangster rap artist.

The chorus for the above song by The Notorious B.I.G. is also strangely prescient:

Not “knowing what they want from me” is a common feeling for probably everyone involved in uptaking AI tooling. We're uncertain of the potential future, but it appears that we are generating more value than ever before. Here's to capturing more of that value for every end user of AI.

The above is my thesis. It's based on Jevons Paradox, which is suggested by Gene Kim and Steve Yegge in their “Vibe Coding” book. We as an industry are still early in figuring out what the exact output value is per dollars spent on tokens. While we're figuring that out, hiring is in a holding pattern. Soon, when we do figure a good equation out for how to best enable developers with AI, we will then start training programs and restructure hierarchies around seniority in this skill set.

There's always more problems to solve and work to do.

If you're reading this, you're going to make it.

You're going to find your way back home on your hero's journey.

[1] “Otherworld” by Bill Xtillidiex Muir is the song playing in the initial scene.