AI Companion 3.0 ushers in hybrid model architecture for secure, scalable, and intelligent enterprise collaboration
Zoom is charting a bold new course in enterprise AI with the launch of AI Companion 3.0, powered by NVIDIA’s Nemotron open technologies. The latest evolution of Zoom’s federated AI architecture introduces a hybrid language model approach that intelligently routes tasks between proprietary Small Language Models (SLMs) and fine-tuned Large Language Models (LLMs), optimizing for latency, cost, and complexity.
At the heart of this transformation is Zoom’s new 49-billion-parameter LLM, developed using NVIDIA’s NeMo tools and Nemotron models. This architecture enables AI Companion to adapt to individual work habits through retrieval-augmented generation (RAG), delivering personalized, real-time support across platforms like Microsoft 365, Google Workspace, Slack, and Salesforce.
“We’ve increased our speed and enhanced lower-cost model decision making.”
—X.D. Huang, CTO, Zoom
“We’ve increased our speed and enhanced lower-cost model decision making using NVIDIA GPUs and AI software stack,” said X.D. Huang, Zoom’s Chief Technology Officer. “This partnership allows us to deliver powerful, security-focused, and scalable AI experiences to our customers at rapid speed.”
Zoom’s federated AI stack dynamically selects the most efficient model for each task—whether it’s transcription, summarization, or multistep reasoning—ensuring high performance and cost efficiency. The integration of NVIDIA’s Llama Nemotron Super-based reasoning model further enhances AI Companion’s ability to handle complex enterprise workflows.
“Our collaboration with Zoom is just beginning.”
—Kari Briski, VP, Generative AI Software, NVIDIA
Kari Briski, Vice President of Generative AI Software at NVIDIA, noted, “The integration of NVIDIA Nemotron into Zoom AI Companion allows enterprises like NVIDIA to experience a private, powerful, and personalized work environment.”
Zoom’s commitment to responsible AI remains central. The company emphasizes that no customer content—audio, video, chat, or documents—is used to train its or third-party AI models. This privacy-first approach positions Zoom to expand its AI capabilities into sensitive sectors like finance, healthcare, and government.
With AI Companion 3.0, Zoom and NVIDIA are not just enhancing productivity—they’re building the infrastructure for the next era of intelligent enterprise collaboration.
