Ep 1 Meet Gar: One Word: Plastics
Episode Highlights
Summary: Pushing a disappointing voice-clone project to the edge reminded me that AI’s real power is amplifying human experience—something I learned decades ago on a factory floor, optimizing seven-second cycles and building worlds with marketing.
The plan that flopped (and why that’s useful)
I spent weeks planning to record three hours of clean, rich audio to train a “nearly indistinguishable” clone of my voice. The documentation promised hyper-realism. My first test? Flat, lifeless, and frankly unusable. I kept iterating, convinced I was doing something wrong—until I realized the failure itself was the lesson. Knowing where AI breaks is as powerful as knowing where it shines.
The unexpected win: talking to myself for hours
What shocked me wasn’t the model—it was me. Once I hit record, I could talk for hours. The next day I wrote in my journal: that was amazing. I’d just experienced how natural it is to externalize the conversations we all have in our heads 24/7. A football coach I know did the same—he sent a 20-minute voice memo that was raw, human, and beautiful. With today’s tools, capturing those monologues and training AI on our lived knowledge might be the most human thing we can do.
Control comes from the edge of failure
I see AI fail a lot. That doesn’t scare me; it empowers me. Pushing the tools until they break is how you stay in control of them. Yes, AI can help you create value and make money—but the better question is: what new, previously impossible things can it let you do for others? Add a little AI-powered value to whatever you already deliver, and you move faster than most because—right now—very few people have a real take.
Flashback: growing up in a plastics factory
Long before prompts and models, I learned about value from molten plastic and seven-second cycle times. As a kid in the 1970s, I cleaned bathrooms, kept machines full, sorted parts, and learned why a customer might say, “I thought God made these.” We were making two million tiny caps a month—the kind that slip over screws under a sink. Every eight seconds: mold closes, injects, opens, ejects. Runners separated. Quality checked. Regrind mixed back in. That’s real-world throughput.
Materials, machines, and messy reality
Materials: polyethylene (friendly), polystyrene (like your soda-cup lid), acetals (strong and… stinky), nylon (great but humidity-sensitive—dry it or suffer brittle parts and steam defects).
Processes: tunnel gates that shear clean; water quenching to prevent warpage; regrind ratios you learn by feel and spec.
Scale: barrels of parts delivered by forklift to my uncle’s packaging shop down the street—he even ran screw-making machines from raw wire.
Becoming “the guy”: operations, then optimization
By my teens I was running weekend shifts solo—turning on machines, grinding scrap, weighing parts, keeping the line humming. Later, I chased automation: barcode tracking, calibrated scales, accounting systems, and my first real computer for CAD so I could help quote jobs and design parts. We invested in new German machines that felt like the BMWs of injection molding. My first big ROI project after graduating NIU (Marketing, 1988): doubling the cap mold from 12 to 24 cavities.
Tiny innovations, outsized impact
A nylon part kept warping when it floated while cooling. The fix? A bubbler—a perforated copper tube with compressed air—to circulate water so every surface cooled evenly. Small idea, huge yield jump. Lesson: network widely, borrow solutions shamelessly.
The Disney blueprint for world-building
My idol growing up was Walt Disney. From multiplane backgrounds and feature-length animation to animatronics and parks, he obsessed over story, sound, transitions, and the feeling of walking through a movie. That level of world-building shaped how I see marketing—and what I do now: craft consistent experiences across content, products, and processes. It’s the same muscle I flexed wanting skylights over sorting tables and plants on the factory floor. Environment matters. Details add up.
AI today: not a bubble—an upgrade
People love saying “AI bubble.” I don’t buy it, because I’ve watched it create real value for clients. The right frame isn’t “AI will make me rich,” it’s “AI lets me deliver more than I could before.” That shift—compounded daily—wins. I still think humans rule, and that’s exactly why we should press AI to its breaking points. At those edges, we learn, we differentiate, and we build things we didn’t think we could.
What the voice-clone saga taught me
The clone disappointed. For now, it’s useless to me.
The recording habit didn’t. Speaking freely for hours—no audience, no rush—created assets I can mine for years.
The meta-lesson: Failures reveal where to lean in. AI is best when it augments distinctly human material—our experiences, voices, and judgment.
What’s next
This backstory introduces a two-part origin series. Part two picks up in 1995, when I left the family business to co-found PlasticsNet—just four years after the first web page went live. After that, the series shifts to more tactical episodes on audience building and our marketing intern process.
