Here's the latest episode of Realm of the Possible We've been testing building fully Open Claw and Claude like agentic functionality into Kilroy by using agentic platforms to give agents the knowledge they need to build apps and modules in Kilroy. Our goal is not to replicate the functionality, but rather to make it superior. In doing so, we've been testing all kinds of llms both centralized and decentralized. Does our theory that breaking memory and knowledge into smaller pieces so that smaller llms can process them still make sense? https://youtube.com/live/-JIOn2mKmrs?feature=share