The Architecture of Trust: Ben Neo on the CIO’s 2026 CX Playbook

The contact centre has long been the industry’s persistent friction point; a high-turnover, high-stress environment where success is often measured by how quickly a customer is ushered off the phone. But as we enter 2026, a fundamental shift is occurring. Zoom, the company that redefined video meetings, is repositioning itself as the architect of a new "cognitive collaboration" layer. At the centre of this transformation is Ben Neo, Head of CX at Zoom.

In this month’s Trailblazer article, we sat down with Ben Neo to address the hard questions facing today’s digital leaders as they navigate the transition from legacy systems to agentic AI.

Ben, looking at 2026, what is the most dangerous assumption a CIO can make when moving from legacy systems to an AI-driven "total experience" strategy?

The most dangerous assumption is the "mushy middle", the belief that you can follow every industry trend simultaneously without choosing a specific architectural risk profile. Many CIOs are trying to be a "Fortress" (building everything in-house for total control) while also trying to be a "Maximiser" (leaning entirely on a single hyperscaler’s ecosystem). You end up with an architecture that is too expensive to be agile, too dependent to be proprietary, and too complex to be stable. In 2026, you must decide: Are you optimising for ownership, scale, or flexibility? Attempting all three is a recipe for a failed transformation.

As we move from AI pilots to full-scale deployment, how should digital leaders shift their focus to ensure they deliver measurable, bottom-line value rather than just "agentic" hype?

You have to stop talking about the tech and start talking about the "outcome-certainty at speed." The hype cycle is over; the board wants to see ROI. We’ve moved the focus from "did the AI answer the question?" to "did the AI resolve the intent?" If your AI is just a faster version of a bad IVR, you haven't delivered value. True value is realised when you empower what I call the "Resolution Specialist", the human agent who is no longer a ticket-solver but a relationship manager, backed by an AI that handles the administrative "work about work" in the background.

David Smith mentions the need for a "Universal Context Layer" to unify data. How can a CIO practically build this foundation without a massive "rip-and-replace" of their existing infrastructure?

You don't need to tear everything down; you need a protocol that acts as a translator. At Zoom, we’re leaning into the Model Context Protocol (MCP). It allows our AI to securely "read the menu" of your existing Jira tickets, Zendesk logs, or OneDrive files without moving the data. By treating your legacy systems as "knowledge bases" rather than obstacles, you create that Universal Context Layer. The AI accesses the context it needs in real -time, respects your existing permissions, and stays grounded in your data. It’s an overlay, not a replacement.

Since fragmented data is the primary cause of AI failure, what are the first three steps an EMEA leader should take to get their "data house" in order?

First, audit for effort, not just satisfaction. Satisfaction is retrospective; effort is predictive. Map where your customers are repeating themselves. That’s where your data is fragmented. Second, move away from point solutions that introduce "siloed intelligence." If your chat AI doesn't talk to your phone AI, your data house is already on fire. Third, adopt a federated AI model. This allows you to swap out LLMs as they evolve while keeping your proprietary "contextual tissue" consistent across the entire enterprise.

Your colleague Dave Michels suggests that the strategic outcome is more important than the "Agentic AI" label. Why is a unified architecture more critical than the individual AI tools a firm chooses?

Tools are transient; architecture is your intellectual property. If you choose a tool based on today’s best LLM, you’re locked into a snapshot of time. A unified architecture, one that connects UCaaS, CCaaS, and AI under a single pane of glass, can help models get smarter and thus your entire organisation gets smarter simultaneously. It’s the difference between buying a faster engine and building a smarter road. The road is what lets you scale.

As AI handles routine tasks with near-zero handle time, which new metrics should CIOs use to prove success to the board?

Average Handle Time (AHT) is dead as a primary metric. In 2026, we track the "AI Resolution Rate", how effectively the AI closes an interaction without any human intervention. We also measure "Effective Escalation". Did the AI pass the customer to the human agent with full context so they didn't have to start over? Finally, we look at "Human Agent Value," which measures things like churn reduction and upsell success on those high-complexity calls that AI didn't take. That’s how you prove the machine is working for the human.

With agents now handling only the most complex cases, how can CIOs use technology to prevent "silent burnout" for these Resolution Specialists?

You can't just throw the hardest problems at humans all day and expect them to stay. We use AI to monitor for "stress signals", sentiment shifts in the conversation or prolonged pauses, and proactively suggest "smart breaks." We also use AI Expert Assist to lower the cognitive load. If the AI is surfacing the right answer in real time, the agent isn't stressed by "searching"; they’re focused on "connecting." We’re using tech to give the agent back their mental bandwidth.

How can EMEA leaders pivot their thinking to treat strict privacy and "sovereign data" as a brand differentiator?

In Europe, sovereignty shouldn't be a compliance hurdle; it’s a trust signal. We’ve built a three-layer framework, Platform, Encryption, and Content, that allows organisations to manage their data according to their specific regional needs. When you can tell a customer, "Your data never leaves this jurisdiction, and you own the keys," that’s a competitive advantage. It moves the conversation from "are you just enabling customers' compliance?" to "are you the most trusted partner?"

Nicolas de Kouchkovsky notes that AI is "tearing down the divide" between sales and service. What is the leadership playbook for unifying these tech stacks?

The playbook starts with a "Shared Memory." Historically, sales saw the "prospect" and service saw the "ticket." In an agentic world, the enterprise has one memory. If a customer mentions a budget constraint to a service agent, that should be context available to the sales team for the next renewal. You unify the stacks by unifying the context layer. Agentic self-service doesn't care which department it’s in; it only cares about the customer journey.

Finally, what is your most vital piece of advice for CIOs to maintain strategic agility when technology is evolving "faster than the speed of light"?

Focus on "Flow," not just "Output." In a world of high-velocity tech, the goal isn't to reach a static "done" state. It’s about building an architecture that allows for constant iteration without friction. Don't build for the AI of today; build for the agility to integrate the AI of tomorrow.

 --