Cerence AI Expands Collaboration with NVIDIA to Advance its CaLLM Family of Language Models

Collaboration with NVIDIA - Cerence Inc announced an expanded collaboration with NVIDIA to enhance its CaLLM family of language models, including the cloud-based Cerence Automotive Large Language Model (CaLLM) and the embedded CaLLM Edge [1] - The collaboration leverages NVIDIA AI Enterprise, a cloud-native software platform, and NVIDIA DRIVE AGX Orin for certain aspects of CaLLM Edge [1] - The integration of agentic frameworks with in-car conversations requires a cross-disciplinary effort combining hardware, software, and UX expertise, with NVIDIA hardware and software engineers contributing to the development [2] Technological Advancements - Cerence AI has accelerated the development and deployment of CaLLM by utilizing NVIDIA AI Enterprise software platform, including NVIDIA TensorRT-LLM and NVIDIA NeMo [2] - The optimization and customization of the CaLLM family of models aim to deliver advanced performance, reduced latency, enhanced privacy and security, and robust protection against malicious interactions [2] - The collaboration enables Cerence AI to develop next-generation user experiences in partnership with automaker customers, enhancing driver experiences [2] Executive Insights - Nils Schanz, Executive Vice President of Product & Technology at Cerence AI, highlighted that optimizing the performance of CaLLM models delivers cost savings and improved performance to automaker customers [3] - Rishi Dhall, Vice President of Automotive at NVIDIA, noted that the collaboration helps Cerence AI deploy advanced NVIDIA AI and accelerated computing technologies to optimize LLM development and deployment [3] Company Overview - Cerence Inc is a global leader in creating AI-powered experiences across automotive and transportation, leveraging decades of innovation in voice, generative AI, and large language models [4] - The company has shipped its technology in over 500 million cars and partners with leading automakers, transportation OEMs, and technology companies to advance user experiences [4] - Cerence is headquartered in Burlington, Massachusetts, with global operations and a worldwide team dedicated to AI innovation [4] Specific Implementations - The collaboration aims to deliver faster in-vehicle assistant performance on NVIDIA accelerated computing and SoCs [5] - Cerence is developing an automotive-optimized implementation of NVIDIA NeMo Guardrails to navigate the nuances of in-car interaction [5] - The implementation and optimization of an agentic architecture on CaLLM Edge via NVIDIA DRIVE AGX Orin will advance the next generation of in-vehicle user experiences [5]