Zendesk announced on Wednesday at the AI summit a series of LLM products intended to reshape the company’s dependence on human technicians.
The center of new features is an autonomous support factor that Zendesk believes will solve 80% of support issues without human intervention. This system will be supplemented by a co-possession who will help human technicians with the remaining 20% of the issues, as well as by a manager, an agent based agent and an analysis agent.
According to Shashi Upadhyay, the president of the product, engineering and AI of Zendesk, the new agents are part of a broader change in the support industry, as AI replaces much of the work previously done by humans.
“People are going to shift from the software made for human people, to a system where AI does most of the project,” Upadhyay told TechCrunch.
Independent benchmarks suggest that modern AI models are capable of undertaking the project. Benchwhich was designed to measure the skill calling the model of a model; Includes a scenario in which models need to process a returning product – a narrow analogue in many support tasks. Today’s leaderClaude Sonnet 4.5, resolves 85% of the issues in the test.
After a chaotic investor race in 2022, Zendesk has made a series of AI acquisitions that laid the groundwork for the current shift. Analytics agent starting today is built directly to the company Acquisition with hyperarccompleted in July. Earlier AI acquisitions include the QA and Agentic Service Klaus (acquired in February 2024) and the Ultimate Automation Platform (acquired next March).
Zendesk has a preview of the new system with existing customers and upadhyay says the results have been promised.
TechCrunch event
Francisco
|
27-29 October 2025
“For customers who have used it, consumer satisfaction has gone from five to ten points,” he told TechCrunch.
Large language models have often been developed for customer support, though rarely on the Zendesk scale. Airbnb companies in regal theatres have already experimented with home support at home, often directly contributing to foundation models. But these systems usually deal with information recovery rather than more complex problems or self -esteem.
If the new impetus for AI support is successful, the financial impacts would be significant. The Zendesk analysis platform already supports about 20,000 customers, resolving 4.6 billion tickets each year. Beyond Zendesk, the US is employed 2.4 million customer service representatives – with much greater workforce in other countries.
