anchorWebex is focused on agentic experiences

AI assistants answer questions. AI agents can go further: they can use approved tools, work with relevant context, and take bounded actions toward a user's goal.

Webex is expanding its AI platform around that agentic model. The goal is to help agents work with trusted collaboration and contact center context while keeping identity, permissions, tool access, and administrator policy in control.

MCP is one way Webex connects agents to tools and data. It is not the whole Webex AI strategy.

anchorMCP is one part of the picture

Start with the interaction your solution needs, then choose the matching connection model.

MCP servers often call REST APIs behind the scenes. A2A connects agents to agents, while MCP connects an agent to capabilities. These models can work together in one solution.

anchorTwo ways MCP connects to Webex

People often call these patterns “inbound MCP” and “outbound MCP,” but those labels change depending on where you stand. The Webex Developer Portal names the two paths Connect Webex MCP Servers to External Clients and Connect External MCP Servers to Webex AI.

Where the agent lives

Tool requests move toward the destination shown in the chart. Results return to the agent through the same connection.

anchorConnect Webex MCP Servers to External Clients

Choose this path when a user or administrator wants an agent in an external AI application to work with Webex.

For example, the agent could:

• Find a meeting and retrieve its transcript.
• Search messages for a project decision.
• Draft or send an update to a Webex space.
• Review calling, workspace, video, or contact center information exposed by an available server.

The Webex MCP server does not contain the agent. It supplies tools to the external agent. The available actions depend on the selected server, the authenticated user's permissions, granted scopes, organization settings, and tool-specific controls.

Start with the Webex MCP server overview, then connect a Webex MCP server to an external client.

anchorConnect External MCP Servers to Webex AI

Choose this path when the Webex AI Agent handling a live customer conversation in Webex Contact Center needs data or an action from a third-party system.

For example, the Webex AI Agent could use an order-processing MCP server to:

• Retrieve the customer's order details and current status.
• Check fulfillment or shipping information.
• Submit an approved order action when the MCP tool supports it.
• Use the result to continue the customer conversation.

In this pattern, the external MCP server does not bring its own agent into Webex Contact Center. It exposes bounded tools to the Webex AI Agent. Developers define those tools, and administrators decide which servers and tools the organization can use.

Start by learning how to configure an MCP client action in Webex AI Agent Studio. To make a server available through the Webex agentic platform, review Agentic Apps concepts and onboard your MCP server.

anchorAgentic Apps provide the governance layer

An Agentic App makes agent-facing capabilities available for discovery and administration in the Webex ecosystem. It can expose capabilities through MCP or, when available, represent an A2A agent.

The developer describes the app and its capabilities. An organization administrator can then review its authentication, enable or block access, select allowed tools, inspect schemas, and reauthorize changes.

Use these guides to continue:

• Onboard an MCP server
• Onboard an A2A agent (beta)
• Provision Agentic Apps in Control Hub
• Review Agentic Apps for administrator approval

anchorChoose your next step