Chewy has rolled out a beta AI customer service assistant called Cai to select mobile app users, and the tool is already resolving nearly one-third of common issues like orders, returns, and account membership without human involvement, CEO Sumit Singh said Wednesday on the company's Q2 2026 earnings call.
The assistant has been live for less than a month and currently reaches under 15% of Chewy's traffic, but the company plans to expand it quickly. The rollout matters for customer support teams because it signals how a major retailer is testing AI that handles routine tickets while routing complex cases to live agents.
What Cai handles today
Cai operates through Chewy's mobile apps and focuses on high-frequency, low-complexity request types. Singh said "shy of one-third" of common issue areas are resolved through its chat capabilities during the beta period.
Chewy has also launched internal AI tools for its live agents. These tools help agents find answers across the various software systems they use, and they speed up onboarding for new hires. Singh said the internal tools are "rapidly allowing us to essentially level those net new agents and their performance much closer to our experienced agents."
For support leaders evaluating AI for Customer Support, Chewy's approach shows a dual-track strategy: customer-facing automation on one side, agent-assist tooling on the other.
A voice AI for veterinary clinics
Chewy has also introduced Callie, a voice-based AI tool for its Chewy Vet Care locations. Callie handles appointment confirmations, scheduling, and routine follow-ups. Singh said the tool reflects "Chewy's brand voice and customers-first tone."
The company expects AI initiatives to produce savings in the low tens of millions of dollars in fiscal 2026, scaling to about $50 million in annual savings by 2027. Singh cited faster agent onboarding as one structural cost reduction, noting that AI tools help offset pressures such as wage inflation.
Where live agents fit
Chewy is not replacing its human support team. Singh said customers using Cai who need or prefer human help "are seamlessly connected to a care team member within seconds."
The company has set a high bar for the assistant's tone and accuracy. "You can expect that the bar that this product has to meet is exceptionally high, and so that's the first design principle," Singh said. "That, from a service bar standpoint, in terms of being true to brand and tone, it has to be spot on."
Chewy shared the AI details alongside Q2 results. Sales rose 7.3% year over year to $3.3 billion, landing on the low end of the company's guidance.
Why this matters for customer support professionals
Chewy's beta data offers a useful benchmark: roughly 30% resolution on common ticket types within the first month, at under 15% traffic exposure. Support teams planning similar rollouts can use that as a reference point for setting expectations with stakeholders. For supervisors managing AI chat assistants and live agent handoffs, the AI Learning Path for Call Center Supervisors covers the operational decisions behind deployments like this one - escalation thresholds, quality monitoring, and agent training when AI handles tier-one volume.
Your membership also unlocks:
