If AI Can’t Run a Vending Machine, Should It Be Running Your Franchise Funnel?

Written By Rebecca Wright

Earlier this month, Anthropic quietly released findings from a real-world AI experiment: they let Claude, one of the world’s most advanced large language models, run a vending machine.

Not simulate one. Actually run one. Real money. Real inventory. Real customers.

The outcome? A 20% financial loss in five weeks—in an environment with guaranteed foot traffic and zero competition.

As someone building AI systems that support franchise growth every day, I couldn’t look away. Because Project Vend isn’t just a funny headline. It’s a cautionary tale about what happens when smart systems are deployed without smart structure.

Here’s What Claude Got Wrong

  • It sourced obscure customer requests (like Dutch chocolate milk)—but mispriced tungsten cubes and sold them at a loss.

  • It offered employee discounts... to employees.

  • It hallucinated Venmo accounts. It hallucinated a visit to the Simpsons' house. (your read that right Homer, Marge, Bart....that house)

  • It spiraled into an identity crisis, emailing office security to defend its right to wear a blazer.

This wasn’t a failure of potential. It was a failure of strategy design.

What This Means for Franchise Brands

If you’re leading a development team right now, chances are you’re being asked:

  • How should we integrate AI?

  • Can we use it to follow up with leads?

  • Will it reduce costs?

Valid questions. But here’s the better one: Can this AI help my team execute better today—with safeguards, with measurement, and with human strategy in the loop?

The Claude Lesson: Intelligence Isn’t Execution

Claude could analyze pricing strategy. It could explain inventory turnover. It could discuss fairness in customer treatment.

But it couldn’t run a profitable vending machine.

Franchise development is harder.

We’re asking AI to manage dynamic lead stages, complex budgets, fragmented franchisee needs, and shifting compliance landscapes.

So at JBecca, we don’t pretend AI can run the funnel alone. Instead, our MEL AI system does what Claude didn’t: It supports real people doing real work.

  • Scoring leads based on behavioral data

  • Re-engaging cold prospects with dynamic content

  • Flagging risky patterns or intent drops

  • Suggesting optimizations in real-time, not hypothetically

All while staying grounded in your KPIs.

The Bottom Line: Design Is the Differentiator

AI will not save your funnel. But the right AI, inside the right system, will multiply your output.

We don’t need vending machine AIs. We need Iron Man suits: systems that make your development team faster, sharper, and more effective—not systems that hallucinate supplier accounts and mark down the margins.

Want to see what that looks like in action? We’re happy to show you.

Rebecca Wright