Google rolled out consumer AI agents in August 2026 that call local stores, check whether a product is in stock, compare prices across locations, and report back with a summary. A second feature monitors prices on tracked items and completes the purchase automatically when they hit a target the user sets. The agents do not answer questions about stores. They do the errand. And that distinction matters more for your business than any model release this year.
If you run a business where customers call, email, or visit to get basic information, Google just trained your customers to expect someone else to do that for them. This technology already works. What you should be thinking about is what happens to your customer experience when the person on the other end of the line is not a person at all.
What Google Actually Shipped
The feature is called “Let Google Call.” When a user searches for a product near them, Google offers to call retailers on their behalf. The agent phones multiple stores, asks specific questions about inventory and pricing, then packages the results into a text or email summary. No human effort required from the consumer’s side after the initial search.
The second feature, agentic checkout, goes further. A user sets a price target on a product. When the price drops to that number, the agent completes the purchase without asking. The consumer does not need to check back, set a reminder, or watch a price tracker. The agent handles the loop from intent to transaction.
As of early August the rollout is still expanding. Google has limited the calling feature to specific product categories. But the pattern is set. An AI agent is making real phone calls to real businesses and returning real answers to real consumers.
Why This Is Not a Consumer Story
The instinct is to file this under “Google shopping feature.” That would be a mistake.
What Google did is normalize a new relationship between consumers and AI. The consumer did not prompt a chatbot. They did not ask for information. They delegated a task and received a completed outcome. That is not search. It is delegation — an agent acting on your behalf in the physical world.
For every business on the receiving end of those calls, the implications land fast. Your front desk, your customer service line, your store associate picking up the phone at 2 PM on a Tuesday is now fielding calls from an AI agent representing a customer who is not there. The customer did not choose to call you. Google’s agent did.
This is the shift that Belief 3 has been pointing at for months. The move from AI-as-tool to AI-as-collaborator does not start with enterprise deployments or developer APIs. It starts with consumer expectations. And consumer expectations flow upstream into every business those consumers interact with.
The Pattern That Matters
Google is not the only one here. Andrew Ng released OpenWorker in late July, an open-source desktop agent that returns finished deliverables instead of chat responses. McKinsey’s CEO confirmed the firm now counts 25,000 AI agents alongside 40,000 human employees, with a goal of reaching parity by year-end (Business Insider, April 2026). Salesforce rebuilt Slackbot from a notification tool into a full agent that searches enterprise data and takes action inside Slack.
The pattern across all of these: agents are crossing the line from “provides information” to “completes work.” The information layer is done. The action layer is here.
Gartner projects 40% of enterprise applications will feature task-specific AI agents by end of 2026, up from under 5% in 2025. That is an 8x increase in one year. But the consumer side is moving faster, because consumers do not have procurement cycles, security reviews, or integration roadmaps. They just use it.
What This Means If You Run a Business
Your customers are being trained right now to expect agents to handle tasks they used to do themselves. Calling a store. Checking availability. Comparing prices. Following up on an order. Every one of those interactions used to require effort from the customer. That effort is being eliminated.
If you treat incoming agent interactions as first-class traffic — structured data, accessible inventory, staff who understand the voice on the phone might not be human — you are positioned for this. You also need to rethink what “customer experience” means when the customer’s agent is the one doing the experiencing.
Keep building for a world where customers pick up the phone themselves, and you will watch the interaction happen without you.
Forget Google’s market share in shopping. This is a behavioral shift that does not reverse. Consumers who discover that an agent can run their errands do not go back to running them manually. The expectation resets. And every business that touches those consumers has to meet the new baseline or watch the interaction happen without them.
Nobody is going to send you a memo about this. Your customers will just stop calling.