The field of conversational AI is moving towards real-time, low-latency interaction, with a focus on enhancing knowledge and semantic understanding. Recent developments have introduced novel hybrid architectures that combine the strengths of speech-to-speech models and large language models, enabling more accurate and informative responses. Additionally, there is a growing interest in enabling efficient communication between large language models, with approaches such as direct semantic communication and selective knowledge sharing. These advancements have the potential to improve the performance and efficiency of multi-agent systems and spoken language models. Noteworthy papers include: KAME, which introduces a tandem architecture for enhancing knowledge in real-time speech-to-speech conversational AI. Cache-to-Cache, which proposes a new paradigm for direct semantic communication between LLMs. KVComm, which enables efficient LLM communication through selective KV sharing. SHANKS, which enables simultaneous hearing and thinking for spoken language models.