Building Rebel Intelligence

It’s 6 a.m. in Cambridge UK, and I’ve been lying awake since 3 a.m., the reason is just because Meta released a new product called Muse, it’s a long-running personal AI assistant inside a confidential VM with long-term memory and the ability to improve the foundational model over time. This is quite similar to the AI system I want and am building now, which makes me excited but also really anxious.

I’m not quite interested with the business model behind Muse, like how they’re going to charge the users, but really impressed by the infrastructure stack they’ve built. Isolated environment for everyone, ensuring privacy and security, a growing memory for better understanding and personalization over time, and the ability to continuously improve the base model itself which loops back into the system, enhancing the overall intelligence. They even announced wearable device integrations in the foreseeable future like the AI glasses which I think it’s the new entrance for the next-generation compute platforms. How could I not be anxious?

And personally, I think Meta is doing the right thing because today they’re behind the frontier AI labs, but they have a big advantage over others as they have built a business empire with billions of users, if they serve their users well, they will not only capture a significant share of the AI market but also in a good position to build better models over time. I also fully believe personal AI or private AI is what everyone needs, that’s why I want to build Rebel Intelligence. The idea is inspired by the rebel alliance in Star Wars, which fights against the oppressive Galactic Empire. It generally means two things:

• Protecting personal data and privacy from centralized tech giants and their control.
• Cultivating intelligence that is deeply personal and distinct, rather than converging into the same.

However, from the technical perspective, what I’m building today is still a bit different from Muse. It’s more like a platform offering the infrastructure for users to build, deploy, and manage their own AI agents and systems rather than the agent itself.

Wrote this down here to clarify my thoughts and keep track of my progress. It’s a long journey ahead, but looks bright.