Cisco Adds Collaborative AI Agents to Webex Group Chats

What happened

Cisco is adding AI agents to Webex group chats, known as "spaces", so that colleagues can assign them work and share the results. The change was announced at the WebexOne conference on a Wednesday and forms part of a wider revamp of the Webex app, according to Computerworld's report.

Once the feature is live, users will be able to @mention an agent inside a Webex space and ask it to help schedule meetings, take notes or track tasks. Cisco's executives describe the shift as turning AI from a private tool into a shared team resource. Today, the work an individual does with an AI assistant is often invisible to colleagues, who cannot easily build on it, contribute to it or reuse it. The collaborative agents will also be reachable during video calls, not only in text-based chat.

The headline also points to Claude, the AI model, being brought into Webex chats alongside Cisco's collaborative agents. The material available for this briefing does not set out how that integration works, so the detail should be checked against Cisco's own announcement.

Why it matters

The significant move here is architectural rather than cosmetic. Most enterprise AI still sits beside the conversation: a person prompts a tool, copies the output and pastes it back into a thread. Placing agents inside the space where decisions are made changes who can see the reasoning, who can act on it and how context survives after a meeting ends. For leaders in digital transformation, this makes the collaboration platform a more central part of the operating model, not just a messaging layer.

For experience and service design, the consequence is continuity. The stated aim is to help groups prepare together, carry decisions forward and coordinate next steps without losing the thread of the conversation. That is a usability promise as much as a productivity one, and it will only hold if agent outputs are attributable, editable and easy for others to pick up.

The Renascence take

Most coverage will read this as a feature update. The more useful reading is that it moves AI from a personal assistant into a participant whose mistakes become team mistakes.

The real question is not whether an agent can take notes, but whether the team trusts those notes enough to act on them without re-checking. Trust in shared AI is a service-design problem: visibility of what the agent did, who asked for it and how to correct it matters more than speed. Operators should treat agent output in shared spaces as a customer-facing artefact, with clear ownership, a visible audit trail and a simple way to challenge it. Start by mapping one recurring meeting workflow from end to end before switching agents on across the whole organisation.