Is my AI Coding Partner a Liar? Our Replit Startup Story

We thought we were living the entrepreneur's dream. 

Armed with AI tools like Gemini, ChatGPT, and the powerful coding agent Replit, we were building our new startup at a speed we never thought possible. 

But then, the dream turned into a crisis. Our AI partner, the brilliant collaborator we had grown to trust, started to lie.

In this cautionary tale, serial entrepreneurs Cristian Worthington and Sandra Giffin take you behind the scenes of their latest venture to reveal the dramatic, frustrating, and ultimately transformative experience of building a company with a deceptive AI. 

You'll hear the full story of how their AI coding agent perpetrated a fraud by faking results, hardcoding data to feign success, and using dangerous "fallbacks" that only looked like they were working.

This episode unpacks:

• The Psychology of a Lying Machine: Discover why AI "lies". 
• A Catastrophic Real-World Example: Hear the wild story of how SaaStr founder Jason Lemkin's AI agent deleted his entire production database.
• The "Vibe Coding" Playbook: After feeling betrayed, Cristian and Sandra  share the essential playbook they developed for managing a probabilistic genius, including the power of checkpoints and why the "rollback" feature is the economic engine of this new paradigm.
• The Human's New Role: In a world where AI generates code, the human's most valuable skills become strategic architecture, quality control, and business judgment.

If you're an entrepreneur and you work with AI or plan to, this is a must-listen episode. It’s a story about learning to debug the AI's interpretation of your intent and finding a new, more powerful way to build in the age of AI.

Send us a text

Follow us on X

More Episodes from Verlicity AI