Zendesk builds autonomous service workforce
- Zendesk on May 19 unveiled an “Autonomous Service Workforce” at its Relate conference, packaging AI agents, copilots and automation as a service operation. - Tom Eggemeier said the “era of the chatbot” is over, while Zendesk said customers will pay only for outcomes it verifiably resolves. - Zendesk said new Agent Builder, omnichannel AI agents and expanded AI capabilities are rolling out through its service platform.
Zendesk used its Relate conference on May 19 to introduce what it calls an “Autonomous Service Workforce,” a product framing that pushes AI in customer support beyond copilots for individual agents and toward department-level operations. The company said the system combines AI agents, copilots, knowledge, actions, governance and analytics inside its service platform. Zendesk also tied the launch to outcome-based pricing, saying customers are billed on issues the company can verify were resolved autonomously. That pitch matters because Zendesk is not presenting the product as another chatbot layer. The company said the workforce model replaces “deflection-based bots” with specialized AI agents that can operate across channels and use cases. In Zendesk’s telling, the unit of sale is no longer a seat or assistant but a system intended to complete service work. (zendesk.com) ### What exactly did Zendesk launch? Zendesk said the launch includes new Agent Builder tools, omnichannel AI agents and copilots tied to what it calls the Zendesk Resolution Platform. In its Relate 2026 product post, the company described the workforce as “a complete, agentic system” designed to run service operations with varying levels of autonomy. (businesswire.com) Zendesk’s help documentation says its more advanced AI agents are designed to handle complex requests with limited direct human supervision, using planning, follow-up questions and authorized actions. The company separately says those agents can work across self-service flows, scripted dialogues and API-connected workflows. (zendesk.com) ### Why is the pricing model central to the launch? Zendesk said its pricing is tied to “verified resolutions,” meaning customers pay when AI agents autonomously resolve issues. The company has been developing that model for some time, but the Relate launch makes it central to the workforce pitch. Tom Eggemeier, Zendesk’s chief executive, said in the company’s announcement that “the era of the chatbot – the era of frustration and deflection – is over.” He said businesses will run on specialized AI agents working alongside human experts and that those agents should be held to standards of accountability similar to human workers. (support.zendesk.com) (zendesk.com) ### How is Zendesk describing the shift from copilots to “workforce”? Zendesk’s own materials describe the platform as combining AI agents, human expertise and automation so service work can move continuously across channels. CMSWire reported the company is billing specialized AI agents on outcomes “the company verifiably resolves,” rather than on seats. (businesswire.com) That language reframes AI from a productivity layer for one support representative into labor capacity for a team. Diginomica described the company’s presentation at Relate as an effort to move from helping support teams manage resolution toward a platform that solves issues directly. That is Diginomica’s characterization, not Zendesk’s wording, but it tracks with the company’s emphasis on autonomous resolution. (businesswire.com) ### What questions does that raise for companies buying it? Zendesk’s own product materials put governance inside the platform, alongside knowledge, actions and insights. The company says customers can grant “as much autonomy as you want,” which implies buyers will need to set boundaries around what agents can decide, when they can take actions, and when work must be escalated to people. (diginomica.com) Zendesk’s support documentation also points buyers to testing, analytics, automated resolution reporting and channel configuration for AI agents. Those features suggest the operational work will not end at deployment. Teams will need people who can configure workflows, monitor outcomes, manage knowledge sources and define escalation paths across customer and employee service environments. That is an inference from the product documentation and rollout materials. (zendesk.com) ### What comes next from here? Zendesk’s support notice says expanded access to its most advanced AI agent capabilities began rolling out on May 11 and is scheduled to run through June 12, 2026. The company’s pricing pages also show AI agents, copilot features and advanced AI agent add-ons across customer-service and employee-service plans, indicating the next phase is commercial rollout rather than a limited concept launch. (support.zendesk.com 1) (support.zendesk.com 2)