Outsourcing My Brain

The 5 Niche, Quirky, and Slightly Embarrassing Ways I Use AI

Look, I’m just as bewildered as you are as to why this custom AI that is specifically trained to draw my face only decided that half of my face would have facial hair. I’m just grateful I got the thing working in the first place.

I've been thinking about how Large Language Models such as ChatGPT or Claude have snuck into my life in ways I didn't anticipate. Not the obvious stuff like "help me write this email to sound less passive-aggressive" — though I do that too — but the weird, specific use cases that probably make me sound like I'm one step away from letting ChatGPT file my taxes.

Which, now that I think about it... wait, no. Let's not give me ideas.

It started innocently enough — a little help with a coding problem here, a quick summary of an article there. But before I knew it, I'd developed a whole ecosystem of AI-assisted workflows that would make my tech-skeptic friends stage an intervention if they knew. (Mom, if you're reading this, I promise I'm still capable of forming my own thoughts. Mostly.)

Here are some niche ways I've been using LLMs that might make you nod in recognition or slowly back away from your screen.

Ridiculous Tech Start-Up Ideas for a Podcast

Look, Minimum Viable Disaster is still a good idea even if we aren't necessarily prolific after releasing our only episode. Is there still something there? Absolutely. Admittedly, the most fun part of the podcast is having Claude generate the most ridiculous tech start-up it could think of. I go with a prompt similar to the following:

Come up with the most ridiculous tech start-up you can think of that somehow involves [INSERT BUZZPHRASE, HOT OBJECT OR POP CULTURE MOMENT HERE]. Pitch me that product like it was a TED talk or Refresh.

You stick that into a conversational AI such as Play.ai or ElevenLabs, and you get something hilarious, like this:

It's so hilariously earnest it could raise $10 million in Silicon Valley tomorrow.

How I Tricked ChatGPT Into Becoming My ADHD Brain Manager

Here's the thing about having ADHD and being obsessed with productivity systems: I've started and abandoned more note-taking methods than I've had hot meals. Have you tried the PARA method? I did. Have you tried Zettelkasten? I got halfway through implementing it before getting distracted by a YouTube video about fermentation. Have you tried bullet journaling? Actually, I stayed with that one for a while and finished an entire paper notebook!

Then, in the middle of my second bullet journal, I remembered I hate writing with a pen and prefer to use a computer for work. (I have no idea where that follow-up bullet journal is nowadays.)

I use AI in MANY, many places in my note-taking.

You all know I use Obsidian as a note-taking tool; a common way to use Obsidian is to write notes that interconnect to each other, like a giant chain of thoughts and imagery similar to what I imagine a brain's thought process would map out to be. One problem, though, is what if you're writing notes on really abstract subjects? What if you are trying to link to something and you just don't know what to link it to?

Then I stumbled on the concept of the “Idea Compass.”

In theory, it's embarrassingly simple: North is upstream (what system is this idea part of?), South is downstream (where are its components?), West connects to reinforcing concepts (what's similar?), and East points to transformative or opposing ideas (what challenges this?). I use this framework for a lot of things now — from structuring my Zettelkasten notes to explaining technical concepts like forward propagation in neural networks. Instead of just creating random links between notes, each connection has a direction that tells me how ideas relate.

Where does AI fit into this spatial madness? When I'm intellectually spent or facing an abstract subject— maybe you can instantly understand what the components of "nihilism" are, but it's not so intuitive for me— LLMs become my cognitive GPS, if you will. I'll ask ChatGPT to "map these concepts according to the Idea Compass," which helps me organize my thoughts using this quirky system. And that's just one example—I've trained it to work with all sorts of productivity frameworks I cycle through during my periodic system-hopping episodes.

When I Made ChatGPT Roleplay as a Confused Junior Engineer (For Educational Purposes)

In my past life as an Enterprise Simulator Leader—which is corporate-speak for "Corporate Trainer" or "person who creates role-play scenarios to train engineers"—I spent countless hours both writing curriculum and crafting realistic technical challenges. These were the kind where you'd pair senior engineers with juniors and have them work through scenarios like "the production database just imploded" or "you deployed on Friday, and now the CEO can't log in." It was normal, fun stuff that definitely didn't give anyone anxiety dreams.

I don't think any of that work saw the light of day, and by contract, I can't share any of that work publicly. Oh well.

How I Outsourced My Conflict Avoidance to an AI (Because That's Totally Healthy)

Let's talk about my most embarrassing use of AI: having ChatGPT draft messages for conversations I'm too chickenshit to have directly. Picture this: it's 11 PM, I'm staring at a text from a friend of a friend about that thing I promised to pick up two weeks ago, or an email from a former colleague who wants to "pick my brain." Instead of responding like a functional adult, I'm hunched over my laptop whispering "help me say no without actually saying no" to a chatbot.

What makes this particularly pathetic—I mean, innovative—is how I've refined the process. I'll give ChatGPT detailed context about why I'm avoiding the interaction, my history with the person, and specific phrases to avoid. "Make it firm but not aggressive, apologetic but not weak, and jesus fucking christ don't use the phrase 'circle back' because that's what I said last time." Then I'll reject its first seven drafts for being "too direct" before finally accepting something so diplomatically padded it barely contains any meaning at all.

Is this emotional outsourcing helping me develop better communication skills? Absolutely not.

But it's preventing me from ghosting people entirely, which feels like the lowest possible bar for personal growth I could set for myself.

The Meta Moment: Having AI Help Me Write About Using AI

The truth is that these AI tools are just extensions of our thoughts—high-tech mirrors that reflect what we put into them, sometimes with surprising clarity and sometimes with hilarious misunderstandings.

What's fascinating is not just how these tools work but how quickly they've become integrated into our creative processes. Two years ago, this kind of workflow would have seemed like science fiction. Now, we call this Thursday.

I don't know if this increasing dependence on AI makes me innovative or just lazy. Maybe both? Either way, I've found these digital assistants are particularly good at amplifying the things I'm already inclined to do—like overthinking simple interactions or finding new and improved ways to avoid finishing projects.

So that's where we are now: collaborating with machines to express how we feel about collaborating with machines. And if that isn't the perfect representation of where we are in 2025, I don't know what is.

And yes, in case you're wondering—I did have Claude help me edit this paragraph. The irony is not lost on me.