Skip to content
‹ Back to IndustriesSoftware & Technology

Cisco invites Claude and OpenAI's dots into Webex meetings, spaces and calls

A WebexOne 2026 demo in Austin previewed Cisco’s push to put AI agents into Webex spaces, meetings and calls. Claude (built on Claude Managed Agents) is described as coming soon, and OpenAI’s dots connects via an MCP integration. Cisco expects its Dialog contact-center harness to reach beta in calendar Q1 2027. Because agents can run for hours or days, IT teams may need to revisit permission boundaries and network capacity assumptions.

This story was produced through MarketScale. See how Software & Technology teams put it to work with Executive Thought Leadership.

By MarketScale Newsroom · CiscoWebexWebexone 2026Ai Agents
Share
Cisco invites Claude and OpenAI's dots into Webex meetings, spaces and calls

Key takeaways

01

An agent that keeps working after the customer hangs up can also keep sending traffic after the office empties, pushing network teams to revisit peak-hour and branch uplink assumptions.

02

Permission scoping is the first pilot question for shared agents: an agent invited into a Webex space should see only what each participant is already authorized to see.

03

Early 2027 is the date to plan around, with Cisco pointing to calendar Q1 2027 for the new Webex experience and for the Dialog contact-center beta.

Free workspace

Turn your Software & Technology expertise into content.

Record interviews, organize footage, and write with AI on a free trial of the MarketScale platform for qualifying companies. No demo required, no credit card.

Try it Free

The refund wasn't allowed. A traveller had missed a flight because of a medical emergency, and the airline's policy said no. On the WebexOne 2026 keynote stage in Austin, the airline's AI agent found the exception process anyway, then made the case to a human escalation specialist using the customer's status and lifetime value. The customer had already left the conversation by then.

That demo, described in Marcus Law's UC Today wrap-up of the event, sums up what Cisco announced in October. AI agents are going into Webex meetings, spaces and calls, into the Webex contact center and into its room devices. Claude, built on Claude Managed Agents, will be able to join a space, meeting or call, pick up the context and carry the follow-up work forward. OpenAI's dots connects through a Model Context Protocol (MCP) integration. Snorre Kjesbu, who runs Cisco Collaboration, described the common thread to UC Today as agents joining the workforce.

The feature list is the easy part. The harder work lands on the people who run Webex and the network underneath it. Agents that keep going for hours or days change two things IT controls: what an agent is allowed to see, and how much traffic a site carries when nobody is in the building.

What an invited agent does in Webex

Cisco's October 7 announcement gives the most concrete picture. In one example, someone on a team can @mention an analytics agent inside a Webex space to retrieve usage statistics, and then have a presentation agent add those figures to a PowerPoint deck. Users can invite agents into spaces, meetings and calls, where they carry out multi-step work across Webex and third-party applications. The dots integration works the other way around. Cisco says the Webex Suite MCP integration feeds Webex context into dots, and that dots will later arrive inside Webex as an always-on personal agent.

Amit Barave, Cisco's VP and GM of Webex Suite and AI, told UC Today that buyers should now judge a collaboration platform on how it moves work along with agents involved and humans in charge. His example of the work involved was compiling a monthly business review. Agents, he said, will run jobs like that for hours or days, overnight if needed.

Timing is still early. The press release says Claude in Webex is coming soon. In an October 12 Techstrong TV interview with Alan Shimel, Craig Burnham, Cisco's vice president of Product and Solutions Marketing, pointed to calendar Q1 2027 for the new Webex experience. He described it as a conversational interface, a daily brief and surfaced priorities, plus the deeper change of letting people and agents work together in meetings and spaces. He also said the first release would keep changing through testing and customer feedback.

Permissions decide whether shared agents work

Barave did not soften his view on governance: "You don't manage them, you govern them." UC Today reported his reasoning. Agents with that much independence are less predictable than conventional software, so IT needs a different kind of control. Cisco is shipping three safeguards with the agents: Webex Compliance Hub with Theta Lake, AI Agent 360, and Pindrop deepfake detection for Webex Meetings and Calling. The press release says Cisco is also adding identity, security, governance and visibility controls. Those controls cover the data AI can reach, the actions it is allowed to take and the way it behaves.

Burnham explained why the controls matter so much in a shared space. The problem Cisco is going after, he told Techstrong TV, is employees running separate AI subscriptions and personal agents whose output stays scattered. Putting that work in one space where colleagues can see results and build on each other's requests only works if the space doesn't leak. Personal information and unrelated projects have to stay out, and each participant should reach only what they're already authorized to use.

That points to where the model could fall short. For organizations whose data permissions are already loosely managed, an agent that draws on shared team context could spread that looseness faster than any person would. Burnham himself treats privacy and trust as the condition that makes Webex AI collaboration useful at all. Firms with clean access boundaries get more out of shared agents, and the rest have cleanup to do first.

In any early pilot, ask Cisco how an agent invited into a meeting or space gets its permissions: from the person who invited it, from the space, or from its own identity. The whole shared-agent model depends on the answer.

Dialog keeps the case open after the call

The airline refund ran on Dialog, a new agentic harness for the Webex AI Agent platform. In a Webex blog post, Vinod Muthukrishnan, VP and GM of Webex Customer Experience, gave the problem a name: the coordination tax, meaning the work customers do repeating information and chasing who owns the next step when teams behind the scenes don't share a view. Dialog keeps one persistent picture of each customer across channels and systems. It's built to keep working after the interaction ends, coordinating with people, agents and systems until the outcome is reached. Cisco expects beta availability in Q1 of calendar 2027.

Two capabilities on Cisco's list matter most to a contact-center leader. Self-learning means improving from journeys, exceptions and human judgment within enterprise policy. Rapid onboarding means starting an agent from the conversations, policies and procedures the business already has. Dialog is also designed as a common layer for Cisco's AI Concierge and AI Receptionist and for third-party conversational agents. Muthukrishnan told CX Today that the first wave of AI agents answered a question and closed a query, and that Dialog is aimed at managing an ongoing relationship across thousands of conversations.

Calling supplies much of that customer context. According to UC Today, more than 82 million users rely on Cisco for communications through Webex Calling, Unified Communications Manager and BroadWorks. Two years ago GetixHealth switched to Webex Contact Center. The company runs revenue cycle operations for hospital chains and provider groups. Dan Womack, its VP of Software, told CX Today that Cisco's roadmap settled the decision. Buyers who choose vendors by roadmap may want to watch Cisco's timing, since the company anticipates beta availability for Dialog in the first quarter of 2027.

There is a network side to this too. In a related Dialog demo, Cisco product marketing director Jessica Smith noted that agents had talked to each other at machine speed to arrange a vacation package while the customer got on with his day. An agent that keeps working on a refund after the customer hangs up is also an agent that keeps sending traffic after the office empties.

An agent that keeps working on a refund after the customer hangs up is also an agent that keeps sending traffic after the office empties.

Zeus Kerravala, writing in Network World, picked out the keynote data point that matters most to network teams. Cisco President and Chief Product Officer Jeetu Patel said chatbot traffic followed human work habits, with spiky peaks. Agents work around the clock and spin up subagents that do the same. He then gave figures on how big the shift is.

How Cisco sizes agent demand

5x
Agent token consumption compared with human consumption, after overtaking it in February, per OpenRouter data Patel cited
450%
More network bandwidth an agent uses for a task than a human doing the same work, by Patel's estimate
60%
Share of global compute Patel expects to go to inference instead of training this year

Network World, reporting Jeetu Patel's WebexOne 2026 keynote (token figure citing OpenRouter)

Kerravala's read is that assumptions about peak hours, oversubscription and branch uplink sizing now need another look, because sustained machine-generated traffic doesn't match the patterns most campus and WAN designs were built for. That makes sense as a direction. The numbers themselves come from a keynote by a company that also sells the network, and the token comparison rests on one routing service's data. They should be checked against local traffic measurement before anyone puts them in a capacity plan.

Conference rooms join the switch port plan

The rooms are changing too. Espen Løberg, Cisco's VP and GM for devices, called collaboration devices the last mile of secure AI infrastructure in the workplace, with AI running on the device. The new Ceiling Speaker Pro brings Cisco's ceiling microphones, cameras and codecs into a fully IP-connected room on a single Ethernet cable, replacing multi-vendor audio setups. Kjesbu's case for it is that IP carries power, media and management together. As Network World points out, that puts PoE budgets, switch port counts and media QoS in charge of gear that used to sit in a separate AV system.

The demand behind the hardware is real. UC Today reports that Cisco Collaboration had its best quarter in seven years in August, up 12% year on year, with video devices up 40%. Workplace Designer now plans whole buildings instead of single rooms. In the demo, it recommended access point counts and placement alongside room devices, because video collaboration and uses such as asset tracking need denser Wi-Fi. According to the press release, Workplace Insights in Cisco Cloud Control compares booked space with actual occupancy, which gives facilities teams their own reason to care.

Ownership is moving as well. Patel told analysts that Cisco Cloud Control, which came together in nine months, is now the single management plane for Cisco products, with Webex Control Hub folded into it. He added that the Meraki and Catalyst brands will eventually go away. For organizations where the collaboration team and the network team still report separately, one console means the line between their jobs will get redrawn whether or not the org chart changes.

The next checkpoint is the first quarter of 2027, when the new Webex experience and the Dialog beta are due. Burnham's comment that the first release will evolve through customer feedback suggests that early pilot sites will shape the permission and traffic controls that everyone else ends up with.

Featured companies

Your experts belong here

Every story in MarketScale Software & Technology starts with a company putting its solutions engineers, product teams, and customer engineers on the record. Buyers are already reading this topic. The only question is whose experts they find.

Buyers ask AI engines who to consider, and published expert answers are what those engines cite.

Book DemoSee how it works15 minutes, straight to a calendar.

About the author

MarketScale Newsroom
MarketScale NewsroomEditorial Team, MarketScale

The MarketScale Newsroom reports on the companies, technologies, and trends shaping 16 B2B industries. It turns primary sources and expert commentary into clear, useful coverage for the people doing the work.

B2B Weekly

The week in Software & Technology, and sixteen other industries, every Monday.

Ten stories, one-line takes, five minutes. Free.

Software & Technology: are you visible to AI?

Before they reach out, Software & Technology buyers ask AI engines which vendors to trust. Explore how your experts, customers, and partners can become useful content for buyers and AI search.

Free Trial

You just read one Software & Technology expert. Your company is full of them.

This article was produced through MarketScale. The same platform turns your solutions engineers, product teams, and customer engineers into the articles, video, and social content Software & Technology buyers are searching for. Start a free trial and see it with your own people. For qualifying companies, no credit card, no demo required.

NPS +73 · 1,000+ creators · 38+ countries

What your free trial includes

Hands-on access to the MarketScale platform
Media requests to your crowd, remote recording, AI writing tools
No demo required. No credit card.
For qualifying companies. Company confirmation required.

More Software & Technology Insights

CREBuilder Connects Brokers to the CommercialCafe Listing Network

CREBuilder connected to the CommercialCafe Listing Network, letting brokers sync listing information from CREBuilder to CommercialCafe, CommercialSearch and PropertyShark. The listing feed carries property details, photos, broker contact information and public marketing materials, and changes made in CREBuilder are reflected through the connection to reduce repeat data entry.

  • 01Brokers can create offering memorandums in CREBuilder and sync listing info to CommercialCafe, CommercialSearch and PropertyShark
  • 02Changes made in CREBuilder are reflected in the listing information through the connection, reducing the need to re-enter listing details across separate systems
  • 03Yardi Listing Network reaches millions of viewers and generates tens of thousands of leads per month

Nov 9, 2026

Stop scaring prospects who already agree on cybersecurity

Stop scaring prospects who already agree on cybersecurity

N-able's Stefanie Hammond applies The JOLT Effect by Matthew Dixon and Ted McKenna to MSP cybersecurity sales. Challenger Inc. summarizes research putting 40% to 60% of lost deals in no decision, and Dixon and McKenna report most no-decision losses stem from fear of failure. Hammond argues MSPs should recommend one program, stop piling on urgency, and phase implementation instead of shrinking protection.

  • 01Challenger Inc.’s February 2024 summary of Dixon and McKenna’s research says 40% to 60% of lost deals end in no decision; in an October 2024 post, the authors wrote that 56% of no-decision losses stem from fear of failure.
  • 02Hammond's rule for MSPs: once a prospect agrees the security gap is real, stop adding fear and breach statistics and start recommending a single program with the reasons behind it.
  • 03Her version of sell less to sell more keeps the full program and splits deployment into monthly phases, identity and MFA first, then endpoint protection and MDR, then backup and recovery testing.

Nov 9, 2026

Musk's Terafab places rush orders with Japanese chip equipment makers

Musk's Terafab places rush orders with Japanese chip equipment makers

Terafab, the chip project jointly pursued by Tesla and SpaceX, has begun ordering from Japanese semiconductor equipment makers, with some delivery deadlines only months away, Nikkei reported. The rush is tied to Elon Musk's target of starting mass production in the second half of 2028. Longer term, Terafab has told equipment makers it intends to reach 1 million wafers per month after 2032, about 700,000 of them for memory.

  • 01Terafab officials visited Japan earlier this month and met suppliers as Terafab began placing orders with equipment makers, some due within months, Nikkei reports.
  • 02Musk is targeting a first test chip in Austin in May or June next year and mass production in the second half of 2028.
  • 03Terafab's long-term goal is 1 million wafers per month after 2032, with about 700,000 for memory and at least 300,000 for logic.

Nov 9, 2026

Explore More Software & Technology Insights

Read more expert perspectives from across Software & Technology.

Browse Software & Technology Hub

About the Expert

MarketScale Newsroom
MarketScale Newsroom

Editorial Team

MarketScale

The MarketScale Newsroom reports on the companies, technologies, and trends shaping 16 B2B industries. It turns primary sources and expert commentary into clear, useful coverage for the people doing the work.

For B2B teams

Your experts could be publishing here

Stories like this one run on content MarketScale captures from real practitioners. See how your team's expertise becomes coverage in Software & Technology and beyond.

Book a Demo

Or call us. No forms required. We pick up. 214-945-2512