Cisco’s unified Webex platform gives it room to pursue agentic collaboration and premium enterprise infrastructure, but its strongest practical advantage may still lie in helping a large calling base modernise on its own terms.

What does Cisco do once it has finished catching up?

That was the question I kept returning to at Cisco WebexOne 2026 in Austin, Texas. Much of the recent Webex story has concerned bringing its capabilities together and connecting collaboration more closely with the wider Cisco business. This year, that groundwork appeared to be giving Cisco room to think further ahead.

The result was an event dominated by agents, context and the infrastructure needed to support them. Cisco wants a larger role in how work happens, extending beyond the communications applications through which people currently organise it.

Yet the most immediately relevant announcement for many customers concerned something considerably more familiar: keeping their existing calling environment while adding cloud intelligence.

That tension runs through Cisco’s direction. It is building for a workplace that may look quite different, while retaining a substantial advantage in helping customers modernise the one they already have.

From catching up to betting ahead

Cisco’s leaders described its platform consolidation as almost complete, still with some outstanding areas and brands such as Meraki and Splunk that will eventually see further technical and marketing integration. It appears that Cisco has caught up on the platform foundations Webex needed. The work of joining capabilities together is now giving way to a much larger bet on what customers might do with them.

Cavell’s coverage of WebexOne 2025 examined the effort to bring Webex into the centre of Cisco’s strategy. This year’s discussions suggested that Cisco is increasingly able to build on that integration. It does not mean every product boundary or customer integration problem has disappeared, but the direction felt more coherent.

The new Webex experience illustrates the change. Cisco presented a personalised homepage intended to help users understand priorities and move work forward, with agents participating across collaboration workflows.

Cisco’s announcement describes Claude Managed Agents as coming soon. It also separates an integration supplying Webex context to OpenAI dots from the future availability of dots inside Webex. These are stages in an announced direction, rather than a single experience that customers can assume is available today.

What felt more convincing was an AI Agent Studio demonstration built around gathering feedback. The agent contacted people, collected responses, and connected with internal resources. The attraction was the reduction in coordination that would otherwise require a series of separate workflows.

This is where collaboration could become more useful: carrying work beyond the conversation. However, somebody still needs to decide whom the agent can contact, which information it can use and what it must never repeat. A successful demonstration establishes a possibility; it does not settle deployment at scale.

Jeetu Patel, Cisco’s President and Chief Product Officer, framed agents, inference, tokens, trust and control as forces shaping the industry. Cisco is making a substantial strategic bet on that world arriving. Cavell’s research suggests many organisations are still working out how to demonstrate value from copilots and assistants. The majority of the market is nowhere near this level of advancement.

Cisco shouldn’t wait until every customer is ready. Thinking ahead is part of its role as an industry leader. The question is whether it can turn an ambitious platform direction into a sequence of useful, manageable purchases for businesses that are several steps behind the keynote.

The infrastructure argument comes at a price

Cisco’s case extends beyond Webex. Its combination of networking, collaboration devices, security and observability makes it well placed to provide the infrastructure large enterprises need as they adopt AI.

Anurag Dhingra’s discussion of context helped explain why. Agents need information from communications, customer systems and documents, but also from the places where people work. Cisco can connect physical workplace information with the network and collaboration environment.

One demonstration made that advantage tangible. A Wi-Fi issue was linked to lunchtime demand in a particular area, with Cisco Spaces supplying building context. The demonstration showed how that understanding could support automated remediation.

There was an identifiable problem, relevant information and a plausible response. That was more persuasive than simply asserting that a broader portfolio must be better.

Device briefings also described more AI processing at the edge, using the compute within Cisco hardware to reduce dependence on central cloud processing. Announced Google Meet support extends the strategy of supporting other collaboration platforms; Cisco’s release schedules RoomOS 27 for November 2026. Similar partnerships and openness with Microsoft Teams support on Cisco devices have already proved extremely fruitful. Precise compatibility and availability still matter, but Cisco can make an infrastructure argument even where Webex is not the customer’s primary collaboration application.

Security provides another connection. In an AI Agent 360 session, Cisco distinguished evaluation, protection against malicious interference and monitoring for an agent drifting in its behaviour. Human involvement remained necessary, including for changes to guardrails. These are practical issues for an organisation delegating work, although visibility alone does not determine responsibility for a mistaken action.

Still, this is a proposition aimed firmly at the upper end of the market. The device discussions repeatedly returned to large enterprises, while calling briefings described Cisco hardware as difficult to fit into parts of the wholesale market. The premium needs to be justified through deployment, support and operating value.

The ceiling-audio proposition offered a useful example: fewer separate components, automatic calibration and unified management could reduce installation and support work. That is a clearer buying argument than asking customers to pay more because the architecture is impressive.

For partners, however, easier deployment may remove work they previously charged for. Integration, adoption and ongoing management could replace that revenue, but Cisco needs to show how. Selling across networking, security and collaboration also requires skills that an established communications partner may not possess.

There is less certainty in how the whole infrastructure story translates into customer experience provision. Security and observability provide credible links; the contribution of room devices to better customer outcomes was less obvious. Cisco has a strong enterprise position, but each part of the platform still needs to earn its place in the customer’s business case.

Voice is still there. It is no longer the main story

Calling felt like a smaller part of Cisco’s overall collaboration narrative. Agents and workplaces occupied more attention, while voice increasingly appeared as something those agents would use. That emphasis left the impression that voice has moved down Cisco’s list of priorities within collaboration.

There are similarities with how Copilot can dominate the conversation around Microsoft. The comparison is about emphasis: AI increasingly defines the story around a communications platform, even when customers still depend on its established functions.

Cisco’s calling leaders pushed back on any suggestion that voice had ceased to matter. In briefings, they insisted it remained a substantial focus. That deserves to sit alongside my observation of the event; reduced prominence does not establish reduced investment. Cisco continues to benefit from the scale of its installed communications base, with more than 80 million telephony users globally across Cisco Unified Communications Manager (CUCM), the BroadSoft platform, and the rapidly growing Webex Calling service. That growth is reflected in markets such as the UK, where Gamma now supports more than 100,000 active users on the Webex Wholesale platform.

A personal voice-agent demonstration showed how the role could change. An agent handled a call while its human user was unavailable, with access to business context. In the demonstrated design, users could specify when it answered and what it did, test the flow, and monitor or join the interaction.

This looked more useful than a voicemail box that simply waits to be checked. Its value will depend on the boundaries: which answers it can provide, when it should escalate, and how much access a person is comfortable delegating.

Briefings placed personal voice agents for Webex Calling in early 2027, and translation preserving a speaker’s voice and tone in 2027. Those remain planned capabilities, rather than evidence of general availability.

Calling leaders also said service providers increasingly wanted to lead with higher-value CX applications. That helps explain why the voice discussion is moving towards assistance and outcomes rather than calling alone.

There is an unresolved tension here. Cisco expects agents to generate more interactions, potentially supporting consumption revenue, yet leadership also acknowledged that agents do not need audio or video to exchange information. More agent activity does not necessarily mean proportionately more voice traffic. Voice remains useful, but its future commercial role is less settled than the demonstrations might suggest.

The calling base may be the more immediate opportunity

Despite that lower prominence, Cisco’s deployment options remain a substantial telephony advantage. Its range of cloud, hybrid and on-premises approaches gives it a strong claim to being the best-placed vendor for customers navigating uncertainty about where communications should run.

Cloud Intelligence for UCM was the clearest expression of that advantage. Cisco’s Webex article describes adding cloud services and AI while retaining Unified Communications Manager call control. Cisco says more than 30 million people still rely on UCM. Bringing that base towards newer capabilities is a different task from persuading every customer to replace its infrastructure.

A hospital with embedded communications dependencies, or a financial institution with specific control requirements, may want AI assistance without rebuilding its calling environment. Cloud Intelligence recognises that innovation and migration can be separate decisions.

The briefings described a phased release without establishing a precise availability schedule for every component. Customers will need that detail before treating the announced package as a deployment plan.

At Cavell, we are seeing sovereignty, regulatory and resilience concerns become more influential, particularly outside the US. Cisco’s options, arguably, fit that discussion better than anyone else in the industry. However, retaining call control on-premises does not mean that all associated AI processing stays there. The location of recordings, transcripts and context still needs scrutiny.

Patel’s observation that Cisco keeps some of its intellectual property outside of AI systems was relevant here. It was not proof that the company distrusts its own providers, but it demonstrated that even an enthusiastic advocate sets boundaries. Customers will need to understand how Cisco’s architecture accommodates theirs.

There is already a narrower, concrete on-premises AI capability. Cisco’s June announcement of general availability for AI PODs for Collaboration specifies local transcription for Cisco Meeting Server, including air-gapped deployment. Summarisation and support for on-premises calling and contact centre appear as future expansion. This is an available capability, with a defined scope, rather than evidence that every agentic function can now run locally.

BroadWorks adds another dimension. Cisco said migration remained slow and discussed adding AI services while partners retain call control. That could let service providers improve the proposition without making migration a condition for progress. Integration requirements, feature support and charges will determine how practical it becomes.

Cisco can pursue an agentic future while recognising that customers need choices about how they reach it. The resulting hybrid environment still has to be managed, but preserving those choices may be more valuable than presenting a single destination.

How far ahead of the customer?

Cisco has caught up on the platform foundations it needed and is now making a much larger bet on what follows. Its breadth supports a credible role in enterprise AI infrastructure, although the premium and delivery demands narrow the immediate audience.

Voice appears to have become less central to the story, but the calling base remains central to the opportunity. Some of Cisco’s most useful innovation may come from helping those customers and partners adopt new capabilities without abandoning what already works.

The question is how closely those two speeds can be aligned. Can Cisco prepare for the workplace it expects while making the next step worthwhile for partners and customers still operating today’s?