Walmart curbs AI assistant usage as costs exceed projections
· Joe Green
Walmart is telling employees to ease up on its internal AI assistant, Code Puppy. The tool, which helps developers write and review code, ended up with way more queries than…
Walmart is telling employees to ease up on its internal AI assistant, Code Puppy. The tool, which helps developers write and review code, ended up with way more queries than Walmart’s systems could comfortably handle. The company had rolled it out with no usage caps or restrictions, but now it’s assigning tiered access to keep the whole thing running without crashing the budget.
Code Puppy is built on a large language model, and employees were encouraged to use it freely. That approach worked fine until demand blew past what Walmart’s infrastructure could support. So now, instead of an open invitation, engineers are being given allocated usage limits. The move is less about pulling back on AI and more about getting the costs under control.
It’s a familiar story for anyone watching enterprise AI rollouts. Companies love the idea of letting everyone experiment with generative AI tools, but the bills add up fast. Running LLMs at scale isn’t cheap, and when usage spikes without a plan, the balance sheet starts to hurt.
What makes this interesting is that Walmart isn’t killing the project. It’s just adding some guardrails. That signals the company still sees real value in Code Puppy, but it’s learning the hard way that you can’t treat AI like an unlimited resource. The takeaway for the rest of the industry might be simple: enthusiasm for AI is great, but you’d better budget for success.