A Bicycle for My Mind:
Lessons in Building and Understanding
As a child, I wanted the freedom of a bike. Today, I'm building an AI tool – another kind of freedom machine.
I used to watch kids flying down the street on their bikes, feeling a mix of admiration and envy. They had freedom to go anywhere, while I trailed behind on foot. Friends would let me borrow their bikes just long enough to taste that freedom. I got good at riding—really good—but those bikes weren't mine, and friends grew tired of sharing.
My father taught me more than how to fix a bike; he showed me how to understand it.
One month before Christmas, my dad took me to a garage sale where we found it: a battered, secondhand bike barely holding itself together. The faded paint, flat tires, and rusted chain initially disappointed me. This wasn't the shiny new bike I'd imagined.
But Dad had a different plan. We took it home and stripped it to the frame. With wire brush, solvent, sandpaper, and LOTS of elbow grease, we transformed it into a pile of shiny clean parts. I had no idea how we'd ever reassemble it. Dad started small—showing me how to set tiny ball bearings and pack them with fresh grease. As we worked, he explained what each part did and why it mattered. It wasn't like a classroom—just the two of us talking, learning as we rebuilt the bike piece by piece.
Just when progress seemed promising, Dad created a Christmas mystery.
Little by little, the bike took shape. We installed new tubes and tires, adjusted the spokes, and tightened every bolt. Then, just days before Christmas, my dad told me I couldn't see it anymore. The last time I had laid eyes on it, the frame was bare metal, the wheels were off, and the handlebars and seat were missing. It looked more like a scattered mess than a bike. I couldn't understand why he had taken it away when so much was left to do.
On Christmas morning, I walked into the living room and froze. The bike I had last seen in pieces—bare metal, wheels missing, looking like a hopeless project—now stood gleaming under the tree. The deep, glossy red paint caught the light, the wheels were straight, seat and handlebars perfectly adjusted. It looked so perfect, I hesitated, wondering if this could really be the same battered frame we had dragged home weeks ago. My dad stood nearby, watching, a knowing smile on his face. "Go on," he said. "Take it for a spin." It was better than any new bike ever.
The real gift wasn't the bike—it was the knowledge to maintain and fix it myself.
Every ride reminded me of the hours spent learning how it all worked. I could fix flats, tighten spokes, adjust brakes. When something broke, I didn't complain—I knew how to make it right. The bike gave me freedom, but the knowledge gave me independence.
Today, I'm building a different kind of machine—an AI tool to extend my thinking.
Years later, I find myself in a similar place with AI. I'm not just trying to use a tool; I'm trying to build one. A bicycle for my mind. Like that old bike, this isn't something that works right out of the box. I've had to strip it down, dig into how it functions, and wrestle with its mechanics.
My goal is simple: reduce the friction of collecting and processing notes.
I'm not a coder, but with IT experience, I'm having AI guide me through the design-build-test process. I'm starting small—using AI to tag, format, and database my notes. Not building the whole bike, just the front wheel.
My AI journey feels like living the same frustrating day on repeat.
Some days, it feels like I'm trapped in a Groundhog Day loop—installing Python, setting up environments, debugging errors, only to start over again. I've built elaborate plans for a system with memory—persistent thoughts that don't vanish between sessions—and hands, a way for AI to take real-world actions. But every time I think I've got it figured out, something breaks. And worse, AI has a habit of confidently handing me hallucinated answers, leading me down rabbit holes that waste entire days before I realize the directions were completely wrong.
It's not just the technical hurdles—it's the nature of AI itself. When I start a session, I have to manually load my Master Preferences Document, a prompt tailored to the task I'm working on, a standards document, and finally, a formatting template. By the time all of that's done, I've often forgotten the topic I meant to write about in the first place.
AI is what I'd call an innocent liar—speaking with total confidence even when wrong.
I learned this lesson the hard way. When I asked AI to help me create persistent storage for templates and preferences, it confidently guided me through building local files, copying documents, and installing Python to run its code. A couple hours of installation, AI-generated coding, setting up an API key, and testing later, I realized something was very wrong. When I finally confronted it, it apologized—none of what we'd built actually let the AI store or recall things. I had spent six hours becoming a human clipboard.
Despite setbacks, I continue because I see the potential.
Like my bike restoration, I know that once I truly understand how this machine works—once I learn to fine-tune and make it my own—it will stop being an obstacle and start being a tool. A tool that extends my reach and allows me to think in new ways.
The wheels are still wobbly, the handlebars crooked, and I'm missing the chain. But I'm learning. And one day, this bicycle for my mind will be ready to ride.
Edited with assistance from Claude 3.7 Sonnet and ChatGPT-4. Illustrations by ChatGPT-4.
That’s My Perspective
One of your best musings!!