Zoom Adds Prompt-Based Agent Builder and Testing Suite to Zoom Virtual Agent

Zoom has introduced Agent Architect and Agent Performance Suite as new capabilities for Zoom Virtual Agent, its customer-service automation product.

According to Zoom’s announcement distributed through Medianet and GlobeNewswire by Notified, Agent Architect is designed to generate “production-ready” voice and digital AI agents from natural-language prompts. The company also announced Agent Performance Suite, a set of tools for testing, validating, measuring, and improving AI agents across their lifecycle.

What Zoom announced

Zoom said Agent Architect allows organizations to create AI agents by describing the desired customer-service experience in prompts. The company positions the feature as a way to reduce manual configuration work when building agents for digital and voice channels.

Zoom’s announcement says the generated agents can be used for customer interactions handled through Zoom Virtual Agent. The company describes the feature as producing agents that are ready for production use, although the announcement does not provide independent performance benchmarks or detailed technical evaluation results.

Agent Performance Suite is aimed at the next stage of deployment: assessing and improving those agents over time. Zoom said the suite can help teams test, validate, optimize, and measure AI agents before and after release. GlobeNewswire’s version of the announcement describes it as supporting the process of testing, measuring, and optimizing agents throughout their lifecycle.

Focus on customer-service reliability

The new tools reflect a broader enterprise concern around AI agents: building them quickly is useful only if they can be evaluated, monitored, and improved reliably. Zoom’s framing emphasizes both creation and ongoing performance management, rather than only agent generation.

CMSWire reported that Zoom is expanding Zoom Virtual Agent with Agent Architect and Agent Performance Suite, and also noted related additions including outcome-based pricing and multi-location deployment capabilities. Those details suggest Zoom is packaging the new agent-building and evaluation features as part of a wider customer-experience push.

Why it matters

For customer-service teams, AI agents are often judged on practical outcomes such as containment, escalation quality, response accuracy, and customer satisfaction. Zoom’s announcement does not publish third-party validation of improvements in those areas, but the company’s focus on testing and optimization addresses a real operational challenge: AI agents require governance after deployment, not just setup at launch.

If Agent Architect works as described, it could shorten the time needed to create voice and digital service agents. If Agent Performance Suite provides meaningful validation and monitoring, it could also help organizations identify weak responses, tune agent behavior, and reduce the risk of poor customer interactions.

The key unanswered questions are how much human review is required before an agent is safe to deploy, what metrics the performance suite exposes, and how Zoom’s tools compare with other contact-center and customer-experience platforms offering AI agent builders and evaluation frameworks.

For now, the sourced announcements establish that Zoom is adding two named capabilities to Zoom Virtual Agent: Agent Architect for prompt-based agent creation, and Agent Performance Suite for testing and ongoing optimization.