Chewy is putting a real number on something most companies only describe in vague terms. The pet supply retailer told investors it expects AI tools to save it 50 million dollars a year, starting in fiscal 2027. It already booked savings in the low tens of millions this past year, and CEO Sumit Singh was careful to tell investors that figure should not be read as pure extra profit, since some of it will offset rising costs elsewhere and some will get reinvested.
That caution is worth paying attention to. Fifty million dollars sounds impressive, but Chewy's annual sales run between roughly 13.46 billion and 13.57 billion dollars. The savings amount to a small slice of revenue. Where it matters more is on the profit line: Chewy is currently running an adjusted profit margin around 6.8 percent and has a long-term target above 10 percent, so an extra 50 million dollars a year is a real, useful push toward that goal, even if it will not get there alone.
The tools behind the number are specific and practical. Kai is a chat assistant inside the Chewy app that handles order status, returns, and subscription changes, and is already resolving about 30 percent of those conversations without a person. Callie is a voice assistant used at some Chewy Vet Care clinics to confirm appointments and book follow-up visits over the phone. AI is also being used to pull information out of pharmacy orders automatically, cutting manual data entry.
What makes Chewy's approach different from a lot of companies chasing the same savings is the groundwork. Before any of these tools went live, Chewy spent several quarters getting its data organized and building what it calls a multiagent orchestration framework, essentially an internal system that lets different AI programs pass tasks to one another instead of working in isolation. It was built in-house rather than assembled from outside software providers. Management believes that gives Chewy a lead that rivals will take years to close, unless they choose to buy the equivalent capability from third parties instead of building it themselves.
The timing lines up with what is happening across the wider economy. Gartner expects total global AI spending to reach 2.59 trillion dollars in 2026, and it expects spending on AI agent software specifically to jump from roughly 206 billion dollars this year to more than 376 billion dollars next year. Separate research from Salesforce found the average company now builds and deploys a new AI agent in under two days, down from about twice that length of time a year earlier, and the average number of agents running per company has nearly tripled over the same period.
Chewy's numbers matter less as a headline and more as a benchmark. It is one of the few companies attaching a specific dollar figure and a specific timeline to AI adoption, in an industry that runs on thin margins and heavy logistics rather than software. Any operator building a similar business case for AI spending now has a real comparison point to work from, instead of a vendor's promise.