Cisco Introduces Dialog AI Framework for Webex Enterprise Platform
The new agentic harness directs AI to manage customer interactions and assist employees with complex tasks.
By Letizia Giugliano
2 min readHuman-Edited, AI-Assisted
Cisco is updating its Webex cloud collaboration platform with a new toolset called Dialog, an "agentic harness" designed to direct artificial intelligence (AI) agents in handling customer interactions. Announced on October 7, 2026, this framework also aims to integrate AI agents seamlessly into employee workflows, though current AI reliability for multi-step tasks presents challenges.
Dialog's Role in Customer Engagement
The Dialog framework is designed to transform fragmented customer interactions into continuous relationships. It tasks various AI agents with maintaining customer satisfaction, allowing them to continue working on a customer's behalf even after a conversation concludes. This system enables AI agents to coordinate across people, other agents, and backend systems to ensure comprehensive support [Engadget].
“The numbers remain dismal. Researchers at Carnegie Mellon University found earlier this year that even the best-performing AI agent, which was Google's Gemini 2.5 Pro at the time, failed to complete real-world office tasks 70 percent of the time.”
— Matthew Noe
Empowering Employees with AI Agents
The updated Webex platform allows any user within a workspace to directly invite AI agents into "spaces, meetings and calls." These agents can execute complex, multi-step work across the Webex platform and integrate with third-party applications. For instance, an AI agent could quickly retrieve statistics during a presentation and update a PowerPoint document in real-time [Engadget].
Agent Onboarding and Performance Refinement
Cisco's new platform aims to speed up the onboarding process for AI agents by treating them "like new teammates." This means agents are given access to existing employee handbooks, knowledge bases, and operating procedures to learn and adapt. The company states that the Dialog platform is intended to "continuously refine agent performance with every interaction," leveraging data analysis from Splunk, a platform Cisco acquired [Engadget].
Current Challenges in AI Agent Reliability
Despite the promises, multiple studies indicate that AI agents are still prone to mistakes, especially in multi-step tasks, which can lead to cascading failures. Researchers at Carnegie Mellon University found that even high-performing AI agents, such as Google's Gemini 2.5 Pro at the time, failed to complete real-world office tasks 70 percent of the time. The failure rate for multi-step tasks generally sits between 60 and 70 percent. For example, an agent with a 95 percent per-step reliability rate would only succeed approximately 36 percent of the time when dealing with a 20-step workflow [Engadget].
What to Watch
While the integration of AI agents into enterprise platforms like Webex offers significant potential for efficiency and improved customer relations, their current reliability for complex tasks remains a critical area for development. The industry anticipates improvements over time, but even a 99 percent per-step reliability rate is considered risky when a company's reputation or financial outcomes are at stake [Engadget].
Read also on The Patent: OpenAI Introduces ChatGPT Health Amid Growing Demand for Health Advice · When evolution meets AI: PROTEUS and the new frontier of molecular design
