WebexOne: Cisco opens collaboration workflows to AI agents

By Marcus Bennett

Cisco's Collaboration business posted its best quarter in seven years in August: growth reached 12% year over year, and video devices rose 40%.

Cisco used WebexOne 2026 in Austin to build on that reported momentum, announcing a plan to put AI agents into customer service, meetings and collaboration hardware. Cisco did not disclose a revenue base or comparison methodology, so the figures do not show what drove growth or whether agents contributed.

Cisco also extended its run as a Leader in Gartner's unified communications as a service (UCaaS) Magic Quadrant, a vendor-positioning framework, to an eighth straight year. The announcement did not provide Gartner's evaluation criteria, and the designation covers a broader category than agent performance.

What does the Austin demo prove?

In the keynote demonstration, a traveler who missed a flight through a medical emergency asked an airline's AI agent for a refund that policy did not allow. The agent found the exception process, then argued the case with a human escalation specialist, citing the traveler's status and lifetime value.

The conversation had already ended, and Cisco did not say whether the refund was approved. Snorre Kjesbu, senior vice president and general manager of Cisco Collaboration, said: “The big idea tying all of this together is that agents are entering the workforce, and that's a major shift.”

Kjesbu said one idea connects the rebuilt Webex App, contact center and room devices. Cisco gave no general-availability dates, adoption figures or rollout scope for that broader portfolio.

How will agents work inside meetings?

Claude, through Claude Managed Agents, will join Webex spaces, meetings and calls, absorb conversational context and carry follow-up work forward. OpenAI's dots will arrive through an MCP (Model Context Protocol) integration. These are announced capabilities; Cisco supplied no accuracy, completion-time, security or pricing benchmarks for either connection.

Amit Barave, vice president and general manager of Webex Suite and AI, said buyers should judge a collaboration platform by how it handles work with agents in the loop, humans in the lead and models in the loop. His operating model puts the agent inside a multi-step process rather than around a single question.

For enterprise collaboration buying teams at pilot stage, the evaluation shifts from chatbot answer quality to context handoff and follow-through. Cisco gave no pilot-team size or test methodology, leaving buyers to demand a controlled trial.

Why must IT govern these agents?

Barave said agents will run multi-step jobs for hours or days, including compiling a monthly business review overnight. He described that independence as less predictable than conventional software and said IT teams need governance, not conventional management. Cisco's control announcement includes:

• AI Agent 360.
• Pindrop deepfake detection in Webex Meetings and Calling.
• Webex Compliance Hub with Theta Lake.

For enterprise IT and procurement teams at deployment approval, the review now needs evidence for each control, not product names alone. Cisco disclosed no test methods, detection rates, deployment scope or pricing.

Microsoft has taken a different route, and Gartner again flagged Copilot as a potentially expensive add-on outside standard UCaaS licensing. No cost estimate or Gartner methodology appeared, while Cisco offered no equivalent agent price, so the comparison signals packaging risk for buyers at contract stage rather than proves which route costs less.

What changes for the contact center?

Dialog, an agentic harness (orchestration layer) for Webex AI Agent, maintains a persistent customer picture across channels and stays with an issue until resolution. Earlier agents typically answered a question and closed a query; Dialog aims to sustain thousands of conversations, hundreds of customer journeys and one continuing relationship.

Vinod Muthukrishnan, vice president and general manager of Webex Customer Experience, said: “Now we are taking a step function up, which is managing relationships: thousands of conversations, hundreds of journeys, but one relationship.” For customer-experience teams at renewal or expansion stage, that makes continuity across channels a procurement test rather than a chatbot feature.

Calling supplies much of this context. Cisco says it powers communications for more than 82 million users across Webex Calling, Unified Communications Manager and BroadWorks; the announcement does not define whether that figure means active users or licensed accounts.

GetixHealth, which handles revenue-cycle work (billing and payment processes) for hospital chains and provider groups, moved to Webex Contact Center two years ago after its previous platform fell behind. Dan Womack, its vice president of software, said Cisco's roadmap decided the choice. The account identifies roadmap confidence as the deciding factor, but provides no deployment size, cost, performance result or comparison with the former platform.

For contact-center buyers running pilots, the next proof points are general-availability dates, licensing, completion and error rates, governance evidence, and independent customer results beyond GetixHealth's selection.