Webex and AGNTCY: How this healthcare booking multi-agent system showcases enterprise innovation

Enterprises are adopting artificial intelligence (AI), but most still struggle with siloed agents that cannot coordinate across vendors, products, or identity boundaries. Crossing this hurdle is vital to achieving what many consider to be the true value of AI: Handling the complex, multi-step workflows that many enterprises rely on to accomplish key outcomes. So, what is holding the industry back? 

A recent survey of over 1,000 U.S.-based enterprise technology professionals found that 86% of organizations need upgrades to their tech stacks to deploy AI agents effectively, citing integration complexity, security, and governance as top blockers. 

These challenges reveal a deeper structural issue: enterprises lack a common foundation for agents to communicate, trust, and operate across diverse environments. Without this foundation, AI agents remain siloed, limited to proprietary frameworks, constrained by fragmented identity systems, and invisible to enterprise controls.  

This is exactly the gap AGNTCY, a Linux Foundation project building the Internet of Agents, was built to address.  

AGNTCY provides an open, interoperable infrastructure that enables agents to discover, authenticate, communicate, and collaborate securely across frameworks and vendors with verifiable identity, unified observability, and policy-based control.  

With AGNTCY, enterprises can build interoperable multi-agent systems (MAS) where agents can find, connect, and coordinate seamlessly across heterogeneous environments while maintaining visibility and control. 

We selected a healthcare contact center as the reference use case, as it demonstrates: 

• Multi-turn interaction: Symptom triage, insurance verification, and scheduling require a secure, low-latency messaging protocol that enables agents to coordinate fluidly in real time.
• Cross-vendor interoperability: Collaboration among agents from Webex, Infermedica, and Infinitus highlights common challenges in discovery, communication, and identity that determine how agents find, recognize, and trust one another across heterogeneous systems.
• Voice workflows: Operating in real-time voice channels creates unique synchronous challenges, requiring end-to-end observability and policy-based governance to ensure compliant, secure handling of patient information.

Healthcare is just one example—but the pattern is broadly applicable to any enterprise workflow where multiple agents need to work together securely and at scale. 

Diagnosis, insurance, and appointment booking multi-agent system with Webex AI agent powered by AGNTCY 

Outshift by Cisco, in partnership with Webex and healthcare AI vendors, Infermedica and Infinitus, developed a healthcare contact center multi-agent system (MAS) demonstration to showcase how AGNTCY enables secure and interoperable collaboration across heterogeneous environments.

Presented at WebexOne 2025, the demo uses the Webex AI Agent to orchestrate multi-agent workflows that connect an intelligent virtual concierge agent with third-party, domain-specific agents to manage multi-turn, real-time voice interactions for symptom triage, insurance verification, and scheduling. 

The demo highlighted how AGNTCY provides the foundational infrastructure layer that includes a common agent directory, trusted identity for agents and MCP servers, protocol integration (A2A and MCP), and end-to-end observability and governance, making the Internet of Agents a practical reality for enterprise-grade, interoperable AI systems.

Here’s how it works: a patient initiates booking through an AI agent that acts as a virtual concierge when deployed to Webex Contact Center, which acts as the supervisor and orchestrator. It consults Infermedica for symptom triage, calls the Infinitus Model Context Protocol (MCP) server to verify insurance coverage, and then uses the Webex Appointment Agent to confirm scheduling. The end goal is to streamline the process for the patient and ensure accuracy for the healthcare teams.

This workflow is powered by AGNTCY’s core components:

• Agent Directory Service makes each agent and MCP server discoverable.
• Agent Identity Service secures interactions with verifiable identity and task-based access controls.
• Observe/Eval SDK provides telemetry and evaluation data to monitor performance.

For builders, this application serves as a reference pattern for orchestrating cross-vendor agents with trust and observability built in.

How to streamline MAS development with AGNTCY architecture 

AGNTCY streamlines MAS development by offering a unified way to discover, secure, and monitor agents:. For this demo, agents built with the AGNTCY Identity SDK and Observe SDK are integrated with the Outshift-hosted Directory SAAs and Identity SaaS services:

• Directory: Developers can onboard agents like Infermedica’s A2A Symptom Triage Agent, Infinitus’ Insurance MCP, and Webex’s Appointment Booking into a shared catalog managed through the Outshift Directory SaaS. 
• Supervisor orchestration: The Webex AI Supervisor agent coordinates workflows using A2A consults and MCP actions via Webex AI Agent Studio, while supporting custom logic from the customer’s own databases.
• Identity and TBAC: Trust is enforced through verifiable agent badges, IdP integration, and task-based access controls (TBAC) powered by the Identity SDK and Identity SaaS, ensuring only authorized actions are executed.
• Observe: The Observe SDK provides runtime telemetry, evaluation schemas, and dashboards for tracking task success, handoff accuracy, and latency.

This creates a repeatable developer workflow: onboard agents with Directory, secure them with Identity, and monitor with Observe—accelerating MAS development in healthcare and extending a similar architecture to finance, insurance, and other industries. 

AGTNCY collaboration: Insights from the Webex, Infermedica, and Infinitus teams

The Outshift and Webex teams worked with Infermedia and Infinitus to show how AGNTCY can transform multi-agent systems across industries. Leaders from each group share their views on the partnership: 

Creating an orchestration framework for multiple AI agents to work with each other has been an important goal for us at Webex AI Agent as our early customers continued to expand their initial scope and use cases for ai agents in customer service. AGNTCY served as the perfect foundation for building A2A (agent-to-agent) collaboration into our product in a secure and scalable fashion.” — Manoj Muthavarapu, Director of Product at Webex

“This collaboration is an important step toward realizing the potential of AI in healthcare—not as isolated tools, but as a coordinated system of agents working together to improve access, efficiency, and outcomes. It’s an inspiring glimpse of what’s possible when innovation is built on shared purpose and interoperability.” —  Piotr Orzechowski, CEO of Infermedica

“Agentic AI is simplifying work and unlocking new possibilities in healthcare. We share Outshift by Cisco’s vision for an Internet of AI agents, one that delivers secure, compliant, and trusted support to patients, clinicians, and staff. With this partnership, we can help accelerate that future for healthcare, and lay foundational standards for open, interoperable, and responsible AI infrastructure.” —  Ankit Jain, CEO and Co-Founder of Infinitus

Start your AGNTCY journey with access to open source MAS 

The Webex AI Agent uses AGNTCY's open source blueprint to enable secure, scalable A2A collaboration, with implementation and decoupling guidance available in the documentation.

We’ve also partnered with Tech Mahindra to develop and open source an alternative voice agent that complements this architecture. This solution demonstrates how Infermedica’s A2A agent and Infinitus’ MCP server integrate seamlessly with AGNTCY components. You’ll be able to explore more details in the GitHub repository.

• Explore the Webex Contact Center AI Solutions 
• Visit the AGNTCY GitHub to use Directory, Identity, and Observe to enforce trust and measure MAS outcomes