Zendesk launches specialised AI agents for customer service

Zendesk has introduced Specialised AI Agents for customer service, including Industry Agents and Custom Agents.

The new agents are designed to understand a business's workflows, knowledge, and connected systems, then take actions across those systems. Zendesk says the tools can automate up to 80% of workflows, initially focusing on more complex service tasks that would otherwise require manual handoffs or staff to move between multiple applications.

The offering has two parts. Industry Agents are built for common tasks within a sector, while Custom Agents are configured by individual businesses around their own policies, approvals, and operating processes.

Zendesk is starting with commerce as its first industry model. In that setting, the agents can help customers shop, manage orders, process returns and exchanges, resolve delivery issues, and handle refunds.

According to Zendesk, the sector-specific agents connect with systems including Shopify, Narvar, Stripe, and Riskified. That lets the software interpret a request, retrieve relevant information, and complete actions within the same workflow, rather than forcing customers or service teams to switch between separate systems.

Zendesk is also working with Riskified to bring risk intelligence into commerce workflows. The company says this could help retailers make decisions on returns, refunds, and exchanges while limiting fraud and policy abuse.

"The era of generic, one-size-fits-all AI agents is over. The best agents are specialists: deeply grounded in the work, the industry, and the business they serve," said Shashi Upadhyay, President of Product, Engineering and AI at Zendesk.

"Specialised Agents give organizations the expertise and context to take on more complex work, earn trust, and drive quantitative impact," Upadhyay said.

Custom workflows

Alongside the industry models, Zendesk is expanding its Agent Builder tool so companies can create and deploy Custom Agents without coding. Businesses can specify the task an agent owns, the systems it may access, the actions it may take, and where human approval is still required.

Customers have already run more than 1 million Custom Agent executions within seven weeks of launch, Zendesk says. Some users have also seen automated resolution rates rise by 10% after deployment.

Zendesk says the agents can also be triggered by business events or run on a schedule, extending automation beyond requests started directly by customers. This allows service teams to use the software for internal processes and follow-up actions, as well as front-line support.

Early users highlighted the tools' operational focus. SodaStream uses custom integrations to adapt service workflows as customer requirements change.

"We want to resolve issues from beginning to end so consumers can get back to using their machines every day," said Leora Machlis, Director of Global Customer Care at SodaStream. "Customizing our integrations helps us act quickly as customer needs change."

Tonies pointed to pressure on response times in consumer support, describing the technology as a way to speed up help for parents dealing with product problems.

"When a Toniebox isn't working, it's not just a technical issue-it's a toddler waiting," said Lauren Cooper, Director of Global Customer Happiness at Tonies. "Parents want answers now, not another hold or a long conversation."

Platform strategy

The launch also reflects a broader effort by Zendesk to position its software across mixed technology estates rather than only within its own platform. Zendesk says Industry Agents and Custom Agents can run not only in Zendesk but also in service environments including Salesforce and ServiceNow.

That cross-platform approach matters for large organizations that have built service operations around several core systems. Rather than requiring a wholesale migration, the agents are intended to work across existing workflows and data sources.

Zendesk says it continuously learns from the outcomes of completed work so organizations can track what succeeded, where agents needed help, and how service processes might improve. The company presents the product as part of its broader push toward what it calls an autonomous service workforce, in which AI takes on a larger share of service operations while companies retain control over rules and approvals.

Riskified, whose tools are being integrated into the commerce offering, said the partnership is intended to add identity and fraud checks to service interactions. That would give retailers a way to assess whether requests linked to refunds, returns, and exchanges are legitimate before action is taken.

"Zendesk Industry Agents are the AI-first platform redefining retail service at scale," said Jeff Otto, Chief Marketing Officer at Riskified. "We're excited to embed our specialist AI agents and Identity Risk Intelligence so that every agent, AI or human, confidently knows who they're actually engaging with, creating delightful experiences for good customers while protecting retailers from fraud and abuse losses."