Cisco adds Dialog AI agents to Webex for multi-step customer support tasks
Cisco is adding Dialog, an agentic AI framework, to its Webex collaboration platform to help manage customer interactions across people, AI agents and backend systems. The change matters because it is designed to let agents continue work after a conversation ends and help employees carry out multi-step tasks across Webex and other applications.
Cisco says Dialog can draw on company handbooks, knowledge bases and procedures when agents are onboarded, and that Splunk will support safety monitoring. Employees will also be able to invite agents into meetings, calls and workspaces; one example is retrieving statistics and updating a PowerPoint during a presentation. The article notes that studies have found AI agents fail 60 to 70 per cent of multi-step tasks, since a mistake at one stage can derail the rest.
- Cisco is bringing its Dialog agentic AI framework to Webex.
- Agents could help with customer interactions and workplace tasks.
- Research cited in the article finds high failure rates for multi-step tasks.
New here? Start with this
Webex is Cisco's collaboration platform, used by organisations for meetings, calls and workspace management. Cisco is now integrating Dialog, a type of artificial intelligence designed to work alongside people and handle tasks that involve multiple steps or decisions.
Traditionally, AI tools have struggled with complex tasks that require remembering where they left off and connecting work across different systems. By adding Dialog to Webex, Cisco aims to let AI agents continue supporting customer interactions even after conversations end, and help employees complete complicated tasks like retrieving data and updating documents during presentations.
Research has shown that AI agents fail roughly 60 to 70 per cent of the time when handling multi-step tasks, because a single mistake early on can cause subsequent steps to fail. Cisco says Dialog can be trained on company handbooks and procedures to reduce these failures, with Splunk providing safety monitoring to catch problems.
Both sides, in good faith
The strongest fair case each way — we don't pick a winner.
The case for
The strong case is that automating multi-step customer support tasks through AI agents can significantly enhance business efficiency and customer experience. By handling routine, multi-stage interactions and administrative work continuously across applications, these systems can free human agents to focus on genuinely complex problems requiring empathy and judgment. With proper safety monitoring through Splunk and access to curated company knowledge bases, Dialog agents can provide consistent, policy-compliant support whilst reducing operational costs—a meaningful advance for businesses managing high-volume customer interactions.
The case against
The legitimate caution is that deploying AI agents for multi-step customer support remains premature given documented reliability problems. Research showing 60-70 per cent failure rates on multi-step tasks presents a genuine risk: customers experiencing cascading failures across multiple sequential actions will likely face frustration, and each mistake compounds the next. Beyond this reliability concern, there are serious questions about job displacement for customer service workers, potential security risks when AI systems access sensitive company knowledge bases, and the complexity of ensuring seamless integration across multiple platforms without single points of failure—risks the announcement does not adequately address.
Read the full article at the source →
Originally published by Engadget as “Cisco is bringing a new agentic framework to its Webex enterprise platform”.