Very interesting to see GoEX the concept of “post-facto LLM validation," and concept of damage confinement, allowing reversible actions and limiting risk in real-world applications. Can't wait to try out the demos