James Stephan-Usypchuk has been building predictive AI systems for 15 years—long before ChatGPT made it mainstream. He's a cancer survivor, serial entrepreneur, and someone who's worked with BlackRock, Blackstone, and private equity firms on deal origination. And he has a psychology degree, which gives him a different lens on the AI hype than most technologists.

In this episode, we dig into the real question organizations are asking right now: AI can do almost anything, but what should it actually do? James makes a sharp distinction between tasks that can be automated and the human elements that can't—and shouldn't—be replaced.

What we cover:

• Why 82% of AI implementations fail when built on bad data
• The difference between using AI as a "magnet in the ocean" versus having it "make sushi on the back of a boat"
• How predictive algorithms can identify who's likely to sell their business—but can't close the deal
• The stat that should scare every marketer: 15,000+ ad impressions per day, but only 3 seconds of recall
• Why the best salespeople he knows are comedians
• What happens when a sales rep is pre-sold and the rep still runs the script

Key insight: "AI can do everything, but so can 50 million other people using AI. When a human steps up with a likability factor you've never seen—automate that. You can't."

Connect with James:

• Website: ecliptica-ops.com
• LinkedIn: James Stephan-Usypchuk