In this episode, we’re back after a long break to talk about where AI is really heading—and why we ended up building our own podcast software from scratch along the way.

We react to a recent clip from Demis Hassabis on the future of large language models and AGI, then share our real-world experience using AI to build complex software without being traditional full-time programmers. From hitting limitations in existing remote podcast tools to creating our own solution that lets us live-share and react to videos together, this episode is about what’s already possible today (not some distant sci-fi future).

We dig into:

• Why AI feels like it’s accelerating again, even without brand-new breakthroughs

• How better memory and continual learning could change everything

• The difference between flashy robot demos and true general intelligence

• And why you don’t need to wait for AGI to start building cool things right now

Most importantly, we show how regular people can use today’s AI tools to create real products and content—like this very podcast.

If you’ve ever thought about starting a podcast, reacting to news clips, or building your own software idea, this is proof you can just start.

Create your own remote podcast, record locally, share your screen, and react to videos live with:

👉 https://podsplice.com

Podsplice is built for simple, high-quality remote recordings with separate tracks, local backups, and easy screen sharing so you can watch and comment on videos together in real time.

Chapters

00:00 – Back After a Year: Why We Had to Build Our Own Software
01:18 – Demis Hassabis on AI Progress and Headroom
03:16 – Do LLMs Need a Big New Breakthrough?
05:03 – Building Complex Software With Plain English Prompts
08:17 – Memory, Personalization, and Continual Learning
11:02 – What “Real” AGI Would Look Like at Home
12:18 – Robots vs Narrow AI: Why Tasks Still Matter
14:14 – Stop Waiting for Perfect AI and Start Creating