Zoom unveils Virtual Agent 3.0 for customer support with AI capabilities to avoid repeat contacts
Zoom has upgraded its virtual AI agent for customer support with new capabilities designed to ensure end-to-end resolution and smoother hand-off to human agents while limiting repeated contacts.
The new Zoom Virtual Agent (ZVA) 3.0 can be used across voice and chat interfaces. It can also interpret documents, forms, images, and serial numbers submitted by customers to extract the relevant information and take actions, automating workflows that previously required manual review, Zoom said in a press release on Wednesday, February 25.
The AI agent’s new capabilities are expected to become generally available over the next few months. It comes at a time when organisations are facing growing pressure to automate more customer service interactions as volumes rise and cost efficiency becomes a priority.
With ZVA 3.0, Zoom said that it is looking to enable its enterprise customers with first-contact resolution, reduced repeat contacts, and end-to-end workflow completion.
It commissioned a recent survey report which found that over 43 per cent of respondents said that chatbots had failed to resolve the issue while 38 per cent said they were frustrated about getting stuck in a loop and another 37 per cent said that they had to repeat information when dealing with customer support chatbots.
“Zoom Virtual Agent 3.0 orchestrates multi-step workflows across systems, continuously learns from human resolutions, and provides full transparency into every agentic action. This allows organizations to confidently automate complex interactions,” Chris Morrissey, general manager of Zoom CX, said in a statement.
ZVA 3.0 is built on top of Zoom’s AI Companion 3.0 architecture. Once known primarily as a video conferencing platform, the B2B cloud-based communications company uses its own AI models as well as LLMs developed by OpenAI and Anthropic to power its AI products and services, including AI agents.
ZVA 3.0 allows customers to automate multi-step workflows involving CRM, billing, order management, and other tasks while ensuring transparency by allowing account admins to see the data sources, decision logic, and workflow paths behind automated actions taken by its AI agent.
CX teams can also perform audits, troubleshoot breakdowns, and refine automation policies, Zoom said.
When integrated with Zoom Contact Centre, its unified platform for customer interactions with enterprise clients, ZVA 3.0 is capable of learning from escalated customer issues that human agents have successfully resolved in the past. It applies these insights and recommendations, with oversight and controls, to similar future requests.
“This creates a structured feedback loop based on human-agent resolutions, reducing repeat contacts and improving resolution consistency over time. The virtual agent can initiate contact, confirm updates, and complete tasks based on known events,” Zoom said.
In terms of performance and accuracy, Zoom said that ZVA 3.0 testing showed the no-match rate drop from 35 per cent to zero. The no-match rate is the percentage of total conversation turns in which a virtual agent failed to understand the user’s input, which means that ZVA 3.0 interpreted almost all customer requests accurately, reducing repeat queries and friction, as per Zoom.
Additionally, Zoom’s billing team saw deflection rates (measuring the percentage of customer support inquiries resolved through self-service tools) rise from 0 per cent to 30 per cent in just three months, saving over 1,000 agent hours per month, the company said.
Comments are closed.