Progress does not come from incremental improvement. It comes from people willing to build what sounds impossible.

In this episode of End of Limits, Pablos Holman breaks down how breakthrough technologies are actually created, using his own experience working with Bill Gates, Jeff Bezos, and now as an investor. He explains why the biggest innovations are rarely solely software, how the majority of the world's industries are ripe for disruption, and why the future belongs to builders who are comfortable operating on a longer time horizon.

This is a conversation about what it really takes to move civilization forward and what we place value in as humans.

We discussed: ◽️ Why radical innovation requires outsiders ◽️ The difference between invention and optimization ◽️ AI’s role in accelerating scientific discovery ◽️ Why most moonshots fail ◽️ The hidden bottlenecks slowing global progress ◽️ How to think like a frontier technologist

Timestamps:

00:00 What real invention actually looks like0

9:10 Why institutions struggle to build breakthrough technology

18:30 AI, robotics, and why LLMs are not autonomy

27:45 Regulation, risk, and outlawing the wrong things

36:20 Open source, competition, and why tech empires fall

45:10 What to build next and where real progress will come from

The future of software is no longer about learning to code. It’s about learning how to build.

In this episode of End of Limits, Replit Founder & CEO Amjad Masad breaks down how AI agents are reshaping software creation, collapsing traditional advantages in tech, and opening product building to an entirely new class of founders.

From agent-driven development to the end of coding as a moat, this conversation explores why speed to execution are becoming the defining constraints of the AI economy.

We discussed:

◽️ The rise of AI agents and realistic timelines for adoption

◽️ What “don’t learn to code, just build” actually means today

◽️ How non-technical founders can now create real products

◽️ Managing AI agents as the next core professional skill

◽️ Creativity, competition, and building in an AI-native world

Timestamps

02:20 The rise of AI agents

14:20 How multi-agent systems collaborate

20:30 Managing agents as a new job skill

30:00 Intelligence, education, and access at scale

39:40 Creativity and the future of building

For podcast-related inquiries: Podcast@humain.com

What if LLMs are only the beginning, and true intelligence is unlocked with world models?

In this episode of End of Limits, Luma Founder & CEO Amit Jain explains why world models and multimodal intelligence are the real next frontier of AI, and why text-only systems will never be enough.

We discussed: ◽️ Why language is only the first step in intelligence ◽️ What “world models” actually mean — and why they matter ◽️ How video teaches AI physics and reality ◽️ Multimodal intelligence beyond text and code ◽️ AI’s impact on creativity, education, and economic growth ◽️ Why intelligence is the true constraint on GDP

00:00 Teaching machines to understand reality 02:40 Why language alone isn’t intelligence 05:15 What world models are 08:30 Why video matters more than text 11:45 How AI learns physics 15:10 Multimodal intelligence explained 18:40 From creativity to robotics 22:30 AI, labor, and economic growth 26:10 Education, tutors, and intelligence at scale 30:00 What comes after LLMs

Podcast-related inquiries, contact us at Podcast@humain.com