To start, this portfolio website was created with Claude Code and Codex. The former was more the muscle, the latter the brain.
So that leaves me, the “owner” of the website, as the monkey typing YES or clicking APPROVE on the screen.
How about that for a start?
My introduction to generative AI was a lot less exciting. Like a lot of people terminally online, I first encountered it through a news article about an AI chatbot by a company called OpenAI. I tried it, wasn’t particularly impressed, and found little use for it back then. Fast forward four years and now, I use it a lot. Them, actually.
I won’t bore you with the minutiae. People swearing by adding AI to their workflows are a dime a dozen online. Suffice to say, my own usage grew gradually, with some discomfort mixed in at first (Spoiler Alert: It doesn’t really go away). Then before I knew it, I was automating tasks that I found terribly tedious and building dashboards that integrate multiple sources of truth. Wait, I’m getting ahead of myself. And, come to think of it… did I really build them?
My discomfort with AI isn’t new, either. I had to contend with it as previous organizations increasingly incorporated it into our workflows (and product lines). My background, both editorial and creative, makes it difficult to be comfortable with the seeming push to remove creative people from, well, creative work. Production work derived solely from generative AI will always leave me with a distaste I find difficult to articulate. And come to think of it, what sort of moat does a business built almost entirely around these tools have when everyone has access to them? What then do you have that sets you apart? Storytelling? A more efficient workflow? More return from dollars/pesos in terms of token spend? Come on.
But I digress.
And perhaps people higher up in the food chain have thought of it and decided that creatives are an acceptable casualty here. The metrics apparently show it, too. From short-form video content made by large production houses to simple social media graphics used by SMEs, generative AI is very much visible. That’s just the visible part of it. I suspect what lies underneath is much larger and more pervasive. And that includes what I’ve been doing these past few months.
I’ve never really considered myself a software developer, and I don’t think I have the experience to call myself one. That’s why my usage of AI had never really touched that apparently larger part of it. I hadn’t even considered the possibility. Until our partners from a foreign multinational visited Manila and had lunch with us.
While lunch was winding down and we were talking about football and trends, the boss mentioned that their company had been using a different AI service provider, but he’d successfully argued for switching the team to Claude. And to prompt usage (hah!), they divided their department into two: the engineers on one side and the marketing people on the other. Their task was to build something. I forgot what. The engineers built theirs the same way they usually built things, from the ground up. The marketing people, despite some initial difficulties, managed to build their own, too. They started with the finished product already in mind and worked their way down, using Claude Code to get there. The finished product changed along the way, of course, but the approach stayed largely the same.
That was all the push I needed to go home that night, get my own Claude subscription, and start exploring what I could do with it. I spent my last months at that company using Claude Code, and later Codex, to automate tedious tasks and build dashboards. There would’ve been a lot more if I weren’t leaving already.
I’ve since left and started my own projects. There’s nothing like seeing my usage allowance reset to a full 100% every Wednesday night and imagining all the things I could build. Before I knew it, I’d acquired an e-ink reader that was surprisingly open to all kinds of development. There was also the thrill of automating tasks for my new clients, showing them what was now possible. As for the e-reader, I’d confess that I’ve spent less time reading on it and more time developing apps for it. Apps that don’t contend with the primary purpose of an e-reader, which is, well, to read.
I had fun with the prototypes, too. Pinned Shop and Pinned Travel, both homages to the original #Pinned project we had at my previous workplace. An e-commerce page that sells pins and a travel website that is not a travel website. How about that.
Then there are the shipped projects. GOTO in the morning, as in porridge, then TOGO in the evening, as in, well, to-go. An automatic news aggregator pulling stories from GMA News (because I apparently couldn’t let go). ON POINT, a departure board showing how many minutes are left before a P2P bus leaves its terminal. I absolutely detested onsite days and commuting on those same P2P buses, but apparently I missed some of it enough to create something like ON POINT. There’s also DOGEAR, a limited weekly reading digest built around literary happenings. And another app idea I’m currently working on.
All of them came from my personal pain points, observations, and use cases. So what if I’m the only one who uses them?
And that leaves us with the question of ownership. What do I really own if all I do is type “Approved, please proceed” all day? Folded into that is the larger issue of AI usage, with all the questions surrounding it. Even with all the personal usage I’ve described, this piece is not a defense of it. If you’re looking for that, perhaps try LinkedIn. Nor is it an indictment, however easy that path could be. It isn’t a black-and-white matter. Not in its entirety, at least.
I cannot claim ownership. At all. There. That’s the gist, basically. With everything created using Claude and Codex, there’s this uneasiness about calling even the most personal projects my own. When people ask, I commit repeated offenses against the English language by using only the passive voice. “It was created with Claude Code.” “Codex was used in developing that.” Skipping the “who” entirely. The initial enthusiasm immediately dissolves into discomfort.
And to be honest, I find the discomfort comforting. That means, at least to me, that despite being accustomed to the tools, I’m still not beyond questioning what they mean. I’d like to think I could walk away from them if I ever had to. However hard that would be, however debilitating to the workflow I’ve created with AI fully enmeshed in it. The discomfort signals that there would be no easy absolution of my complicity.
Although I suspect there will come a time when I’ll simply become comfortable with it. The blurred lines between creator, owner, and developer. AI usage will eventually become so ubiquitous that perhaps the distinctions I’m struggling with today will no longer seem particularly meaningful. The larger issues surrounding AI will remain, of course, particularly for people working in creative fields. Getting used to the tools doesn’t make those questions go away.
But that time is not now. And I’m okay with it.
For now, the monkey will keep on typing. Eventually, they might reproduce Shakespeare by repeatedly banging their head against a keyboard.
Or not Shakespeare, even. That’s coming from a monkey who has read some of his works and was left thoroughly underwhelmed.