Agentic Virtual Agent: New Features for a Better Customer Experience

Agent-Based Customer Experience Orchestration

Genesys, a leading global provider of agent-based customer experience (CX) orchestration solutions, continues to develop Genesys Cloud Agentic Virtual Agent. This enables companies to automate more complex customer interactions faster, more reliably, and with more natural conversations. Thanks to its integration with enterprise systems and enhanced interoperability, the Agentic Virtual Agent can also collaborate more securely with business systems and other AI agents to optimize customer outcomes across the entire enterprise. Since its general availability in March, ongoing enhancements have made it increasingly easier to create, test, and optimize the Agentic Virtual Agent in conjunction with Scaled Cognition’s new large-scale action model, APT-2, as well as with new native language capabilities and integrations with ElevenLabs and Deepgram. In addition, the solution benefits from deeper integration with the Model Context Protocol (MCP) as well as Agent2Agent (A2A) interoperability with enterprise platforms.

More and more companies are relying on the Agentic Virtual Agent to go beyond traditional automation

, including Global Payments, MTN South Africa, Utility Warehouse, UOL, and Charles Sturt University. For some companies, this is already translating into savings of tens of millions per year. Customers also report increased satisfaction and a rise in the Net Promoter Score of more than 70 points. Some deployments were up and running after just a few weeks.

Greater reliability for automated customer experiences

Thanks to Scaled Cognition’s latest large-scale action model, APT-2, Genesys is further enhancing the intelligence behind Agentic Virtual Agent. This model is designed to ensure more accurate conclusions, greater fidelity to the facts, and more reliable execution. APT-2 supports more comprehensive agent specifications, longer and more complex workflows, faster response times, and improved performance across multiple languages. These improvements help Agentic Virtual Agent analyze increasingly complex customer requests and see them through to resolution, while adhering to the limits defined by businesses. At the same time, they help prevent errors in the customer experience that could undermine customer trust.

Designing World-Class Voice Conversations

Agentic Virtual Agent’s new voice features ensure more natural and responsive conversations. Thanks to End-of-Turn Detection, the AI agent recognizes when a customer has finished speaking—making conversations flow more smoothly, with fewer awkward pauses and interruptions. With Contextual Tuning, the Agentic Virtual Agent also automatically adapts terminology, tone, and style to the specific company, its brand image, and the context of the interaction. Furthermore, through new integrations with ElevenLabs and Deepgram in the areas of text-to-speech and conversational speech recognition, Genesys is expanding the range of options available to customers.

• Thanks to ElevenLabs, companies have access to expressive, natural-sounding voices for engaging conversations with the virtual assistant. At the same time, Genesys Cloud’s governance and control features remain fully intact.
• Deepgram provides real-time speech recognition designed for dialogue-oriented AI, which enables the Agentic Virtual Agent to detect interruptions, changes in topic, and shifts in customer intent throughout the conversation. This richer, more up-to-date context provides the agent with a better foundation for drawing conclusions and taking action, enabling it to respond more quickly and complete tasks more reliably.

These new features make voice dialogs with the Agentic Virtual Agent smoother and more natural. This enables companies to create engaging customer experiences that increase self-service success rates and reduce unnecessary transfers to human agents.

Build Trustworthy Agentic Experiences Faster

Throughout the development cycle, companies can create and deploy experiences with the Agentic Virtual Agent more quickly. Thanks to specification-based development, teams can describe customer experiences and desired outcomes using existing documents such as process documentation, standard operating procedures, and conversation transcripts; AI also helps translate these specifications into ready-to-use configurations. Additionally, development teams can leverage various tools such as Anthropic’s Claude Code, OpenAI Codex, Cursor, and Kiro to create specifications that can then be integrated into Genesys Cloud for testing and governance—providing greater flexibility while ensuring consistent control across the enterprise. In addition, Genesys is launching a new AI-powered design tool that analyzes the design of virtual assistants, identifies opportunities for efficiency improvements, and provides actionable recommendations for better customer outcomes. New AI-powered testing procedures, advanced evaluation features, and detailed traceability also help companies validate Agentic virtual agents before deployment, understand their actions, and continuously improve their performance.

Creating the Most Connected Agentic Virtual Agent on the Market

Leveraging orchestration capabilities from the recent acquisition of Pinkfish, Genesys enhances the Agentic Virtual Agent with more than 500 connections to enterprise systems as well as access to over 25,000 MCP-compatible tools —making it much easier to integrate the AI agent into the business context as well as the systems and tools underlying customer interactions. It can thus securely access CRM, ERP, IT service management, knowledge base, collaboration, billing, and other business applications to understand the context and resolve customer requests.

Greater Capability Through Collaboration Among Multiple AI Agents

Thanks to A2A interoperability, the Agentic Virtual Agent can securely collaborate with specialized AI agents on platforms such as Salesforce and ServiceNow, delegate tasks, and coordinate work steps across multi-level business processes. Combined with MCP-based access to enterprise tools, A2A links customer conversations to the systems and AI agents specifically needed to resolve each request. For example, the Agentic Virtual Agent can coordinate with specialized AI agents on various enterprise platforms to resolve a billing issue—from reviewing account information to verifying policies and performing background tasks— all while maintaining context throughout the interaction with the customer. Additionally, the Agentic Virtual Agent is available for customer communication via digital channels such as WhatsApp. This enables companies to deliver “Agentic” customer experiences through customers’ preferred channels, while A2A and MCP solutions remain interoperable with underlying enterprise systems and AI agents.

Customer Experiences Around the World

Around the world, companies are already achieving measurable business success with the Agentic Virtual Agent, including:

• Riachuelo, one of Brazil’s largest fashion retailers, replaced its traditional chatbot with the Genesys Agentic Virtual Agent, thereby increasing customer retention from 30% to 84%, improving the CSAT score by 60 points in the first year, and boosting productivity by more than 400%—without hiring additional staff.
• Zeppelin Rental GmbH, one of the leading providers of rental and construction services, is already enhancing its digital customer experience with Genesys Agentic Virtual Agent. Since customers can describe their construction site needs in natural language—rather than browsing through product catalogs—they can more easily find the relevant products, services, and information. The result is a simpler and more intuitive self-service experience that helps relieve pressure on traditional service channels.

“Enterprise AI is entering a new phase—one that is measured first and foremost by what it delivers to customers”

says Mike Szilagyi, senior vice president and head of product at Genesys. “We’re equipping the Genesys Cloud Agentic virtual agent with more context, intelligence, and capabilities so it can operate at an enterprise scale, collaborate with other AI agents, and resolve increasingly complex customer inquiries. This enables companies to turn customer requests into meaningful outcomes at scale.”