Zoom CTO XD Huang on Bringing Steve Jobs to Every Meeting
XD Huang spent more than 30 years at Microsoft before joining Zoom. On the TNW podcast, he explains why Zoom runs several AI models simultaneously, likening it to a panel of gymnastics judges.
October 6, 2026 - 10:24 am
Zoom CTO XD Huang (left) and TNW’s Cristian Dina in Amsterdam, 22 September 2026 Credit: TNW
Zoom has opened the speech recognition behind its My Notes tool to other companies, and is in conversation with some AI note-taking firms considering switching to it, said XD Huang, Zoom's chief technology officer. He explained that Zoom aims for AI that completes meeting discussions before the meeting ends.
Huang leads Zoom’s AI strategy. Prior to Zoom, he spent three decades at Microsoft, where he was a technical fellow and CTO of Azure AI, contributing to the launch of the Azure OpenAI Service. His teams achieved human-level parity in speech recognition, machine translation, natural language understanding, and computer vision. He holds over 170 US patents and is a member of the US National Academy of Engineering.
Huang spoke with Cristian Dina for the second episode of the newly launched TNW Podcast, recorded on September 22nd in Amsterdam at a Zoom event preceding HumanX. The inaugural episode featured Yahoo’s Jim Lanzone. The full conversation is available below.
“Just imagine the possibility. You can afford to bring Steve Jobs to every meeting you are attending,” Huang said.
Huang refers to this shift as “C2C”, transitioning from conversation to completion. The initial C focuses on Zoom's established meeting business, predominantly during the pandemic. The second C involves AI completing tasks on a user’s behalf, with their authorization. He predicts 2026 as “the year of completion”.
My Notes captures conversations in Zoom, Microsoft, and Google meetings, as well as in-person, according to Huang, who added that many users are unaware it functions outside Zoom.
Opening Up the Scribe API
Startups developed bots to join calls and generate summaries, prompting Dina to inquire if Zoom had rendered them obsolete with My Notes.
“I wouldn’t say we crushed [them]. And actually, we open this up,” Huang responded.
The technology behind My Notes is accessible to everyone through the Scribe API, part of Zoom AI Services. Scribe has consistently topped the Open ASR Leaderboard on Hugging Face since its initial release, which benchmarks over 60 speech recognition systems. Version two is now available.
This API powers ZoomMate, Zoom’s AI assistant, and its customer support tools. Integrating with Slack, Jira, and Salesforce, it allows search across Zoom, Google Drive, and Microsoft documents, providing context.
A Federated Approach to AI
Dina inquired about calls for slowing AI development. Huang emphasized Zoom's focus on safety, security, and compliance, and the company's reliance on a federated approach, where multiple models scrutinize each other to minimize hallucinations and enhance accuracy.
“When you actually play gymnastics, you know why there are multiple judges,” he said.
On Humanity’s Last Exam, a test of expert knowledge, combining several frontier models outperformed each individual model. Zoom utilizes models from OpenAI and Anthropic, alongside its own tailored small language model. Complex tasks are handled by smaller models, while the most challenging assignments go to frontier models.
Huang noted that Zoom developed its speech recognition and translation internally. Its meeting summarization model rivals ChatGPT at a significantly lower cost because it is tailored for this specific task.
A Journey Through Speech Recognition
Born in China and educated in Scotland, Xuedong Huang initiated his work on speech recognition in 1982 as a graduate student in China.