Dan Lambert believes robotics is finally reaching an inflection point. Robots are getting cheaper, NVIDIA’s newest hardware is making real-time physical AI possible, and AI coding tools are making robotics development accessible to far more people.

But the software layer in between is still painfully difficult.

After spending more than 10 hours setting up a robot using existing tools, Dan started building Peppy to reduce that process to something closer to 10 minutes. The company is creating a prompt-based robotics platform that lets developers describe what they want a robot to do, test it in simulation, and eventually deploy that behavior to physical hardware.

In Episode 002 of First Slice, Dan joins Jared S. Taylor to explain why he thinks robotics is about to have its “vibe coding” moment, what is still holding general-purpose robots back, and why the industry needs far more real-world failures before robots become truly useful.

They also discuss:

• Why sub-$5,000 robots could dramatically expand who can build with robotics
• Peppy’s goal of turning a 10-hour robotics setup process into roughly 10 minutes
• Why Dan thinks robots may eventually be upgraded and replaced like smartphones
• The shortage of real-world physical data compared with the massive datasets used to train LLMs
• The “sim-to-real gap” and why robots need to get “punched in the face” by real-world failures
• Peppy’s roughly $1 million oversubscribed pre-seed round
• How a plain-English prompt gets translated into actual robot movement
• Why AI hallucinations become much more dangerous when software controls physical machines
• Peppy’s safety layer for preventing AI-generated code from damaging robots or people
• Whether robots should ever be allowed to use knives and other potentially dangerous objects
• Why NVIDIA has become foundational infrastructure for modern robotics
• Peppy’s positioning as a “Lovable for Robotics”
• The idea of getting a robotics application 90% of the way there in simulation before buying hardware
• Peppy’s longer-term vision of becoming an Android-like software and app layer across different robot bodies
• Dan’s belief that millions of robots need to enter the real world before the industry has enough failure data to make major progress
• Peppy’s planned public launch in December 2026
• The moment Dan first became obsessed with robotics and why he thinks humans are approaching a historic technological shift

Matthew Lenhard discovered an underground market selling access to AI models from companies like OpenAI, Anthropic and DeepSeek at massive discounts.

The problem became impossible for him to ignore after seeing companies hit with more than $1 million in token fraud per day, including one incident totaling roughly $10 million in 48 hours.

A few days later, he quit his job and started building Vectoral.

In Episode 001 of First Slice, Matthew joins Jared S. Taylor to explain how the underground AI token market works, who supplies the tokens, who buys them, and why AI is creating entirely new forms of software abuse.

They also discuss:

• AI inference being sold for up to 98% below official pricing
• Matthew contacting token brokers and being offered roughly $3 million per month of OpenAI inference
• “Denial of wallet” attacks that intentionally run up a company's AI bill
• How Vectoral detects abusive AI usage
• Why fraud may move from the major AI labs to the applications built on top of them
• Matthew's prediction that hijacking AI agents could become a major future attack vector
• How he decides which problems are worth building companies around