I read a lot. Articles, newsletters, TikToks, podcasts — I'm constantly pulling in ideas. When something matches the vision I have in my head, I save it, share it, try to make sense of it.
Here's my problem: I've always been able to see things clearly in my head. The full picture. How everything connects. But translating that into something real — a product, a document, a thing that exists in the world — that's where I'd get stuck. The gap between vision and execution was frustrating.
That's changing. The tools exist now to bridge that gap. And this week, a few things clicked that I want to share.
The Spark
I saw a TikTok from @byericf — just a quick clip captioned "Jarvis, here we come." It's about building AI assistants. The Iron Man dream.
The concepts in that video weren't new to me. I'd been thinking about this stuff for years. But putting it into words? That was hard. Watching that video, I thought: that's what I've been trying to say.
The gap between "full Jarvis" and "fill out this web form" is where the alpha lives.
We're not getting Jarvis tomorrow. But we don't have to keep building software like it's 2015 either. There's a middle ground — and that's what I'm building.
What I Want Providers to Feel
I've been designing the Upcycle provider experience. The app that haulers will use every day to manage their jobs, communicate with customers, track their earnings.
Most work apps feel like tools you have to wrestle with. Dashboards with a dozen tabs. Settings buried three clicks deep. Information scattered everywhere.
I want something different.
When a provider opens Upcycle at 6am, coffee in hand, I want them to feel like they have a talented coworker. A trusted assistant who's already thought about their day. Someone who says: "Here's what's on deck. First job's in Elk Grove, heavy load — bring the trailer. Traffic looks good. Ready to roll?"
Not a dashboard. A partner.
That's the difference between software that works for you and software that makes you work.
The Tech Exists
I went deep on the tooling this week. Turns out, smart people have been building toward this.
The version that makes sense for production: you define a library of components (cards, maps, charts), and the AI picks from your list based on context. Safe. Controllable. And it feels like magic.
Narrative Engineering
I read a piece this week from Erica Wenger at Park Rangers Capital about "narrative engineering." The idea that the best startups don't hire storytellers — they're founded by people who engineer the conditions for belief to compound.
Companies that hire for narrative without founder-level conviction end up with content calendars. Companies where it's in the founding DNA end up with movements.
Read that again. Content calendars vs. movements.
The ideas started flowing as I read that post. It's exactly what I was trying to articulate. I could see how it all fits together in my head — and now the tools exist that can bring that vision into reality.
