Back office ops · Production
Zoom leverages AssemblyAI Speech AI models to advance Zoom AI Companion research and development
The problem
Zoom sought to advance their speech-to-text research and development for Zoom AI Companion, aiming to improve transcription accuracy and differentiate from competitors.
Workflow diagram · grounded in source
1
AssemblyAI models selected for R&D
trigger
“AssemblyAI's industry-leading Speech AI models were selected to help advance Zoom's research and development efforts around speech-to-text”
2
Refine AI Companion training data
ai_action
“using these models to refine data used to train Zoom's AI Companion, strengthening Zoom's ability to deliver high-performance AI features”
3
Rare word accuracy enhancement
ai_action
“AssemblyAI's high accuracy in rare words like product names, people names, and other proper nouns, as well as uniquely formatted phrases, to enhance Zoom's own model capabilities in recognizing those words for better transcription quality”
4
Enhanced meeting intelligence output
output
“More precise meeting summaries that capture the nuances of complex discussions. Reliable action item tracking with accurate speaker attribution. Enhanced searchability of meeting content, including content shared in a meeting and meeting…”
Reported outcome
By including AssemblyAI's Speech AI models in their R&D, Zoom expects significant advancement in making AI-powered collaboration tools more accurate and reliable, including more precise meeting summaries and improved transcription.
Reported metrics
advancement in AI collaboration toolssignificant advancement
Reported stack
AssemblyAIZoom AI Companion
Frequently asked questions
What did this team achieve with this AI workflow?
By including AssemblyAI's Speech AI models in their R&D, Zoom expects significant advancement in making AI-powered collaboration tools more accurate and reliable, including more precise meeting summaries and improved…
What tools did this team use?
AssemblyAI, Zoom AI Companion.
What results were reported?
advancement in AI collaboration tools: significant advancement (source-reported, not independently verified).
How is this back office ops AI workflow structured?
AssemblyAI models selected for R&D → Refine AI Companion training data → Rare word accuracy enhancement → Enhanced meeting intelligence output.