We filmed this episode of Data Pour at People’s Market in Myers Park, Charlotte—grabbed drinks, hit record, and got into the kind of conversation that happens when two people who grew up in “big data” start comparing notes on where AI is actually headed.

Funny thing is—this bottle shop closed down a week after we filmed. Perfect metaphor for tech, honestly: everything feels stable until it isn’t.

Why James

James Barney is one of my closest friends and mentors. I’ve known him for a long time, and he’s been the person I go to when I want an AI take that isn’t hype and isn’t fear—just reality.

He’s spent the last decade-plus inside big enterprises, helping teams evaluate what’s real, what scales, and what’s just noise. Right now, he’s helping lead AI initiatives at a major insurance company—meaning his day job is basically: everyone wants the next big thing… which parts of this are going to stick?

Our shared roots: fintech, big data, and invisible systems

We both started our tech careers in big data, and a lot of our growth came from the same kind of pressure cooker—fintech needs.

If you’ve worked in regulated financial environments, you know the deal: huge volumes of data, constant scrutiny, and systems that have to be accurate, auditable, and fast. That world forces you to get good at the unsexy stuff—digesting, categorizing, transforming, and moving data reliably.

In the episode, we talk about the era when we were building Kafka infrastructure before half the managed conveniences existed. It was messy, duct-tape engineering—but it’s also the foundation for how so many modern “invisible” systems work today.

The point: AI is new… but the patterns aren’t

One of the threads we pull on: AI feels new to everyone, but the underlying enterprise challenges rhyme with what we already lived through with cloud and big data.

The model is only part of the story. The real value shows up when you add:

  • clean, governed data
  • real context
  • integrations into systems that matter
  • tools that can retrieve or act (not just “generate text”)

We also get into why copilots can feel underwhelming in enterprises: the public tools people use every day are “fully layered.” Inside a company, you’re often starting with the plain base layer—and you have to build the rest.

Watch the full episode

This blog is just the background and the framing—if you want the full story (and the full vibe), watch the YouTube episode. It’s a real conversation between two people who’ve been in the trenches, trying to describe what’s actually happening in AI without turning it into marketing copy or doomposting.

Link to the episode: https://youtu.be/ibtZckIQ_JI?si=XiD_CPVBWdinvugr