I didn’t attend the WebexOne event in person this week, but the livestream keynote and announcements make it clear Cisco is reading the same market signals we are. The future of collaboration isn’t just a better chat window. It’s a shared workspace for humans and digital coworkers.
I’ve been tracking how the enterprise is responding to the rise of shadow AI. Employees are bringing zero-UI personal agents to work, and the only way to stop that data drain is by providing a safer alternative natively in the corporate stack. Cisco just joined Google and Salesforce in trying to build that alternative, but their approach leans heavily on what they already do best. They’re focusing on network level observability and contact center flow.
Their main move was introducing Claude Managed Agents. Through a partnership with Anthropic, Claude now sits inside Webex spaces and calls as a full team member. It’s built for long-running, multi-step work that continues overnight after the human team logs off. It’s essentially their take on the digital coworker. And just like Google, Cisco is refusing to lock into a single model. They announced Model Context Protocol (MCP) integrations with OpenAI and a path for organizations to host third-party models locally for data sovereignty.
Where Cisco’s architecture attempts to differentiate is on memory and governance. They introduced Webex Dialog as an agentic harness designed to maintain a persistent customer view across different channels and teams. This fits exactly into how a Universal Context Layer should operate. The vendor that holds an organization’s work history and customer context graph effectively owns the enterprise switching costs because you can’t easily rip out an agent that knows your entire history.
But you can’t have agents running autonomous loops without a way to monitor them. That’s where Splunk comes in. Cisco is embedding Agent Observability with Splunk into their AI Agent 360 environment. Cisco calls this AgenticOps, giving IT teams a shared view of agent health, escalation patterns, and real-time token spend.
The real test for Cisco will be whether AgenticOps can bridge the gap between performance tracking and legal proof. Regulators and CISOs don’t just want operational dashboards showing how a system is running. They want the digital equivalent of a locked and tamper-proof receipt for every action an agent takes. If Cisco can turn Splunk’s monitoring data into an unalterable audit trail, they’ll solve one of the biggest friction points in enterprise AI.
It’s actually a distinct architectural swing. Salesforce is tying agents to CRM data, and Google’s building a horizontal employee across the cloud. Cisco is betting that putting agents inside the collaboration layer and backing them with Splunk will give enterprises the confidence to finally let AI run in the background. If they can execute, it’s another validation of Continuous Flow Architecture as the market continues to shift.
