Cisco is updating its Webex collaboration platform with a new agentic framework called Dialog. The company describes Dialog as an "agentic harness" that coordinates AI agents across people, other agents, and backend systems. Its stated goal is to turn "fragmented interactions into continuous customer relationships," with agents continuing to work on a customer's behalf after a live conversation ends.
Dialog also aims to speed up AI onboarding by treating agents like new teammates, letting them review employee handbooks, knowledge bases, and operating procedures. For employees, the updated Webex will allow anyone to invite AI agents directly into spaces, meetings, and calls to execute complex, multi-step work. Cisco gives the example of an agent pulling statistics during a presentation and updating a PowerPoint on the fly. The company says the platform, which uses its Splunk data analysis tools, continuously refines agent performance with every interaction.
However, Engadget notes that the reliability of AI agents remains a serious open question. It cites a Carnegie Mellon study from earlier this year finding that even the best-performing agent at the time, Google's Gemini 2.5 Pro, failed to complete real-world office tasks 70 percent of the time. Broader research cited in the article puts agent failure rates for multi-step tasks between 60 and 70 percent, because a single mistake in one step can derail the entire workflow. Even an agent with 95 percent per-step reliability would succeed only about 36 percent of the time on a 20-step task, suggesting that Cisco's promises may take time to prove out in practice.