AI is forcing a rethink of the agent workspace. As service organizations move from simple copilots to autonomous AI agents, the old model of tabs, side panels, disconnected apps, and swivel-chair workflows is starting to show its age.
The issue is practical. AI agents need context, permissions, workflows, and knowledge to resolve work safely. Human agents need the same context when an interaction escalates.
When that environment is fragmented, both sides struggle. For Hannah Deveney, Senior Director, Product Management at Zendesk, the shift has made older service architecture a growing operational risk:
“Relying on that more traditional best-of-breed architecture from the past is a strategic vulnerability now and really blocking the capability that you can take forward with AI.”
That means agentic AI is becoming part of the service workforce.
Deveney highlighted that Gartner predicts by 2029, agentic AI will autonomously resolve 80% of common customer service issues without human intervention, while reducing operational costs by 30%. That creates a new question for CX leaders: can their current workspace support that kind of operating model?
Fragmented Workspaces Create Context Blind Spots
Contact centers have lived with fragmented systems for years. Agents jump between CRM records, telephony tools, ticket histories, knowledge bases, order systems, and channel-specific interfaces. Many have become skilled at working around that complexity.
AI makes those workarounds more exposed. Deveney argued that fragmented service stacks create a “context blind spot” for AI agents. An AI agent may see the support ticket, but miss the customer’s recent purchase, promotional eligibility, service history, or prior escalation.
That can turn a potentially useful AI interaction into a frustrating one. Deveney described the impact of that fractured environment:
“A fragmented tech stack creates bottlenecks of data, which doesn’t lead to a great customer experience or customer resolution.”
The customer expectation is already clear. Zendesk’s CX statistics show that 70% of customers expect anyone they interact with to have the full context of their situation.
That expectation applies whether the customer reaches a human agent, an AI agent, or a blend of both.
When the context breaks, the customer does not usually care which system caused the issue. They experience the same outcome: repetition, delay, and a service journey that feels disconnected.
The Sidebar Problem Is Now an AI Problem
The app sidebar used to be a useful compromise. It allowed contact center teams to extend existing systems without rebuilding the entire workspace. It helped agents access tools, context, and channel controls from alongside the ticket. Yet AI-led service raises the bar.
If an AI agent suggests a refund, updates a shipping address, or escalates a billing issue, the human agent cannot be pushed into another tab to complete the work. They need to approve, adjust, or act in the same flow. Deveney made this point clearly:
“Agents shouldn’t need to leave the conversation to take action.”
That is also the logic behind Zendesk’s native Contact Center update. In June 2026, Zendesk announced the general availability of its native Contact Center, built directly into the Zendesk Agent Workspace.
The announcement moved call controls into the workspace without a separate Marketplace app, introduced single status management across Zendesk and Amazon Connect, simplified authentication, and gave admins native Contact Center settings from the Product menu.
Zendesk said the change came from customer feedback. The previous sidebar app felt disconnected, obscured the ticket workspace, and made Contact Center feel like a third-party application rather than a native part of Zendesk.
That product update reflects a wider market reality. The agent workspace is no longer an afterthought. It is part of the service architecture.
Handoffs Need Complete Context
The hardest moments in AI-led service often happen at handoff. An AI agent may handle the initial triage, gather data, and attempt the first resolution path. But when the issue hits a complexity threshold, requires empathy, or needs approval, a human agent has to step in. That handoff cannot feel like a restart.
The handoff between an AI agent and a human agent is where a fragmented workspace becomes especially visible. A customer should not have to repeat their issue, and the employee taking over should be able to understand the situation quickly enough to keep the conversation moving.
That requires a shared view of the interaction, including the customer’s intent, relevant account context, and the reason the issue needs human attention. The escalation may be triggered by a policy exception, a missing piece of information, or a point where human judgement is needed.