Zendesk says its new AI agent can solve 80% of support issues

Zendesk announced Wednesday at its AI summit a string of LLM-driven products meant to reshape the company’s reliance on human technicians.

The center of the new features is an autonomous support agent that Zendesk believes will solve 80% of support issues without human intervention. That system will be supplemented by a co-pilot agent, which will assist human technicians with the remaining 20% of issues, as well as an admin-layer agent, a voice-based agent, and an analytics agent.

According to Shashi Upadhyay, Zendesk’s president of Product, Engineering and AI, the new agents are part of a broader change in the support industry, as AI replaces much of the work that was previously done by humans.

“The world’s going to shift from software that’s built for human users, to a system where AI actually does most of the work,” Upadhyay told TechCrunch.

Independent benchmarks suggest that contemporary AI models are capable of taking on the work. TAU-bench, which was designed to measure a model’s tool-calling ability, includes a scenario in which models have to process a returned product — a close analogue to many support tasks. The current leader, Claude Sonnet 4.5, solves 85% of issues on the test.

After a chaotic investor fight in 2022, Zendesk has made a string of AI acquisitions that laid the groundwork for the current shift. The analytics agent launching today is built directly on the company’s Hyperarc acquisition, which was completed in July. Earlier AI acquisitions include the QA and agentic service system Klaus (acquired in February 2024) and the automation platform Ultimate (acquired the following March).

Zendesk has previewed the new system with existing customers, and Upadhyay says the results have been promising.

“For customers that have been using it, consumer satisfaction has been up by five to 10 points,” he told TechCrunch.

Large language models have often been deployed for customer support, although rarely at the Zendesk’s scale. Companies from Airbnb to Regal Theaters have already experimented with in-house chatbot support, often contracting directly with foundation model labs. But those systems typically deal with information retrieval rather than more complex troubleshooting or taking self-directed action.

If the new push for AI-based support is successful, the economic implications would be significant. Zendesk’s Resolution Platform already supports nearly 20,000 customers, resolving 4.6 billion tickets each year. Beyond Zendesk, the U.S. employs 2.4 million customer service representatives — with far larger workforces in other countries.