2 weeks Ago
Zoom is transforming from a video conferencing platform into an AI-powered productivity hub through ...
More pioneering agentic AI systems that can reason, plan, and act autonomously to solve workplace challenges. In the bustling world of artificial intelligence, where buzzwords come and go faster than Silicon Valley startups, one term is genuinely reshaping how we think about work: agentic AI. While most companies are still figuring out how to implement basic generative AI tools, Zoom is already several steps ahead, quietly building AI systems that don't just respond to commands but actively anticipate needs, reason through complex problems, and take autonomous action.
During a recent conversation with XD Huang, Zoom's Chief Technology Officer, I discovered how the company famous for keeping us connected during the pandemic is now spearheading a workplace revolution that few saw coming. According to Huang, agentic AI represents the fulfillment of a dream that tech visionaries have pursued for generations. "Software has been maturing, and intelligence has been the dream for many generations," he explains.
"Finally, with the evolution of generative AI large language models, the capability reached the stage that this is the beginning of the application of agentic AI." But what distinguishes genuinely agentic systems from the increasingly crowded AI landscape? Huang defines four essential characteristics: reasoning capability, long-term memory, proactive and autonomous behavior, and task orchestration abilities. "The ability to autonomously identify the opportunity and the context and the task to help you to execute, of course, with your consent, will be essentially very important," Huang tells me.
"This is a complex task, usually requires multiple steps, multiple agents. We need the ability to orchestrate all of them together to complete tasks that are beyond just a few minutes' work." While many still think of Zoom primarily as a video conferencing platform, the company has been undergoing a significant transformation.
"Zoom is really transforming from a video conferencing company that is well known in every country to an AI-first productivity company," Huang states. At the center of this transformation is Zoom's AI Companion, which Huang describes as your "chief of staff." This AI assistant exemplifies Zoom's new vision, summarized elegantly by Huang: "We are transforming from meeting to milestone.
The reason we're having meetings is we want to achieve our next milestone." This evolution makes perfect sense. Meetings aren't ultimately about conversation; they're about accomplishing objectives.
By focusing on this fundamental truth, Zoom is reorienting its entire product philosophy around outcomes rather than activities. What exactly can Zoom's AI Companion do today? Far more than you might expect. During meetings, it doesn't just record and transcribe conversations – it actively listens, reasons based on context, and identifies follow-up tasks.
Huang describes how it assigns tasks to specific team members and, increasingly, can even execute those tasks autonomously. In one particularly telling example, Huang shared a personal experience: "I received a piece of email. I need to meet with three people but without the slot.
I can just ask AI Companion to schedule a meeting with my opening slot. AI Companion will just do that on my behalf." Even more impressively, Zoom's AI capabilities now extend beyond virtual meetings.
"I attended one of the customer advisory board meetings. There were like 70 people in the room, and I just turned on AI Companion, left my iPhone on the table, and after the meeting, I got the meeting minutes the same as I got like a virtual meeting." Perhaps most surprisingly, Huang revealed how he used AI Companion to draft his own performance review.
"AI Companion can access all the documents I created. So that is the knowledge work I have, all the decisions I made, all the recommendations I had..
. AI Companion, I would say, fulfilled 80% of the task." While much of the industry obsesses over ever-larger language models, Zoom has taken a fascinating counter-approach by developing specialized small language models (SLMs).
Specifically, they've created a 2 billion parameter model that achieves remarkable efficiency. Huang explains the strategy using an apt analogy: "The analogy I would say is like we have encyclopedia that got everything, but we also have specialty. We go very deep.
The society needs both." This approach reflects what Huang calls "T-shaped knowledge" – broad horizontal understanding paired with deep vertical expertise. The results have been impressive.
"The quality of this 2 billion small language model, the performance on the general benchmark is as good as the larger language model that was two years ago," Huang states. When customized for specific tasks like translation, these smaller models can actually outperform much larger competitors while costing a fraction to run. "In the future, I believe that everyone will probably have a number of agents, maybe 10 or 20.
Each agent can be equipped with a small language model that's optimized for that agent," Huang predicts. "And the larger language model will bring those small language models working together, orchestrate them seamlessly." Despite his enthusiasm for AI's potential, Huang emphasizes that humans remain central to Zoom's vision.
The company isn't aiming to replace human collaboration but to dramatically enhance it. "We need to have personal AI Companion, AI Companion to learn your way of doing business. How much do you want to control?" This learning process mirrors the real-world relationship between executives and their support staff.
"If any task requires your approval, your chief of staff in the first month may not know, but over time, your chief of staff will get used to that," Huang explains. "We'll actually make a good decision, that is the journey between the human and AI working together." When asked how agentic AI will transform workplace collaboration over the next few years, Huang returns to his core theme: "From meeting to milestone.
Your milestone can be far more advanced because of the productivity, you can be more ambitious." He paints a picture of AI assistants that monitor relevant information around the clock, conduct research autonomously, and return with recommendations – all while keeping humans in control. "Imagine you'll have a hundred people working on your behalf.
After this meeting, they will actually analyze what we talked about and go out to really do research, collect information, and come back with the best recommendation possible." Zoom's transformation offers valuable lessons for every business. While the company could have rested on its pandemic-fueled success, it has instead reimagined its entire purpose around a more fundamental human need: turning conversations into accomplishments.
As Huang succinctly puts it: "Conversation is a human nature. That is not gonna change. We are still going to converse, and we are still going to speak.
But the milestone and the completion, the bar will be much, much higher than what we have today." With AI increasingly capable of handling routine tasks, we're free to set our sights higher – on bigger milestones and more ambitious goals. The future of work isn't just about smarter technology; it's about heightened human ambition made possible by intelligent automation.
As I reflect on my conversation with Huang, I'm struck by how Zoom has positioned itself at the intersection of AI capability and human aspiration. In the process, it may have discovered the perfect formula for workplace technology: tools that don't just help us communicate better but help us achieve more..
