How would cloud-native actually help you on building multimodal AI in production? What is cloud-native anyway? Is it just a buzzword to get VC's money yet delivers an empty promise?