Online meetings used to be much the same as their physical world counterparts.
With the introduction of generative AI, online meeting platforms began to add new insights, including voice transcription services. Platform vendors, including Microsoft Teams, Google Meet and Cisco WebEx, have steadily integrated capabilities that go beyond what in-person physical meetings can provide.
Now in the emerging era of agentic AI, online meetings are poised to diverge even further with a new wave of innovations. Zoom recently announced its agentic AI efforts which aims to create the paradigm shift from meetings to milestones. Microsoft has added copilot actions which can integrate with its Microsoft Teams service helping users inside of meetings to get insight and connect with other Microsoft services. Cisco has been steadily expanding its AI capabilities in Webex, and announced a Webex AI agent at the end of 2024 that helps with contact center deployments.
Another firm that has been particularly active in the space is Otter AI, which is somewhat differentiated in that it is not directly tethered to any specific online meeting vendor platform. While Otter first made its mark as an AI-powered voice transcription service, it has added an AI assistant called Otter Pilot, an AI chat assistant, and a series of meeting capabilities known as Meeting GenAI.
Today, Otter is going a step further, with its foray into agentic AI. While some of the agentic AI features that Otter is adding are not unique, it is doing at least one thing that isn’t yet part of every agentic AI meeting technology. Otter AI is now being integrated as an entity inside of a meeting that can actually respond by voice to queries.
No longer is the AI an external participant accessible just via a chat window; AI is now a live entity that is literally part of the meeting.
The rise of AI meeting agents: Beyond silent observers
For the last several years, AI meeting assistants have been passive observers—transcribing conversations, creating summaries and allowing post-meeting queries. Otter is now changing this dynamic with its AI Meeting Agent, which can actively participate in conversations when summoned.
“The new AI meeting agent we’re building will be able to help you with voice in real-time meetings,” Sam Liang, CEO of Otter AI told VentureBeat. “During the meeting, you can say, ‘Hey Otter,’ and ask it questions.”
In a live demonstration with VentureBeat, Liang showed how the agent could answer factual questions, provide meeting summaries and even schedule follow-up meetings—all through voice commands during an active conversation.
What makes this particularly powerful is the agent’s ability to connect to a company’s knowledge ecosystem.
“This agent can become a domain expert,” Liang noted. “When you’re having a meeting, it has almost infinite knowledge from the internet, but this agent also has knowledge about your enterprise.”
Autonomous agents: When AI runs the meeting
Taking agentic capability a step further, Otter is also launching an autonomous SDR (Sales Development Representative) agent that can independently conduct entire meetings without human intervention. This agent greets website visitors, conducts product demonstrations and schedules follow-up meetings with human sales representatives.
“We cannot hire a million human agents to answer questions, but we built this Otter SDR agent that functions like a sales development representative who can greet every single visitor and give them a live demo,” Liang said.
The use of chatbots and avatar-based systems is not new. Liang argued that what distinguishes his company’s technology from existing avatar-based solutions is its ability to conduct multimedia product demonstrations in real time. The autonomous agent can share screens, demonstrate product features and respond to specific questions about functionality and pricing.
The technical architecture of agentic AI for meetings
Agentic AI is an overloaded and somewhat overhyped term in the industry today overall.
Functionally agentic AI is about enabling actions, which can be done by combining multiple models with a tool like LangChain or by using the function-calling capabilities present in many models.
Liang has an even more nuanced definition for agentic AI.
“Agents in general are a more sophisticated AI system that can break down a large and complicated task into smaller tasks,” he said. “It can do reasoning and it can do some planning to perform a task.”
Otter is not using LangChain but has developed its own custom technology specifically designed for the challenges of multi-speaker voice environments. The technical architecture combines both public knowledge retrieval and proprietary enterprise information through a custom RAG (Retrieval-Augmented Generation) implementation. This enables the agent to understand company-specific information like employee names, project terminology and internal acronyms.
The future of meeting intelligence isn’t just agentic AI
What agentic AI is bringing to online meeting platforms represents a powerful new set of capabilities for organizational efficiency.
For decades, organizational efficiency experts have warned about the risks of wasted time in meetings. Modern AI-powered platforms are changing that risk. Last week Zoom’s CTO told me that his goal was to move the technology from meetings to milestones, where the output of a meeting isn’t just another meeting but actionable things that will benefit the organization.
While agentic AI can create workflows, there is still also benefit in regular AI assistants that are not actually agentic. There will still be standalone AI assistants and fully agentic ones and that’s a good thing, according to Anurag Dhingra, SVP & GM, Enterprise Connectivity and Collaboration at Cisco.
“While AI agents act as autonomous do-ers and AI assistants serve as prompted helpers, both offer benefits in boosting productivity and enhancing overall collaboration,” Dhingra told VentureBeat. “It’s not a matter of choosing one over the other but rather leveraging their combined strengths to create environments where teams can focus on innovation and strategic decision-making.”
What is also starting to happen is more interoperability across different platforms. For example, Cisco’s AI Assistant will work with workflow applications such as Salesforce, ServiceNow and Outlook.
Strategic implications of Agentic AI for enterprise meetings
For enterprises evaluating their AI adoption roadmap, Otter’s approach to meeting agents represents a strategic inflection point.
Rather than implementing general-purpose AI and hoping for ROI, organizations should consider how domain-specific meeting agents can address concrete pain points with measurable impacts.
As AI continues to evolve from tools we use, to colleagues we work with, the differentiation will increasingly be found not in the underlying models but in how effectively they’re trained to understand specific business contexts and domain knowledge.
The post Agentic AI is changing online meeting platforms: Moving from silent observer to active participant appeared first on Venture Beat.
