Yelp's AI Agent Now Calls Restaurants to Check Wait Times

Yelp's fall product release, published this morning, put a voice agent inside one of the most mundane consumer chores: finding out how long the wait is at a restaurant. The feature is called Check for Me. It appears on restaurant pages in the iOS app, where a user enters party size and a phone number. The agent then places an actual phone call to the restaurant and texts back the answer within a few minutes, according to TechCrunch.

A deliberately narrow agent

Check for Me is a single-purpose agent, and that is precisely why it has a chance of working. The task has a fixed script: ask the host for the current wait time for a given party size, capture the number, return it as a text message. No browsing, no open-ended persuasion, no payment.

Yelp has sold a voice product to restaurants and service businesses since last year. This release adds the consumer side of the same call, which means Yelp now sits on both ends of the conversation, as the company's product chief framed it. The rest of the release widens the assistant: an OpenTable integration so reservations happen without leaving the app, memory for preferences from earlier chats, 30 days of chat history, scheduled reminders, and shareable My Lists.

The agent-to-agent risk

TechCrunch flagged the structural problem the feature walks into: a user's agent may end up talking to the restaurant's agent. Host-side AI receptionists are increasingly common, and two agents optimized for brevity can stall on a question as simple as how long the wait is. There is no shared handshake, no reliable way for the calling agent to detect it is speaking to a bot, and no defined escalation path to a human.

That failure mode is not Yelp-specific. It is already showing up in enterprise workflows where an inbound qualification agent meets an outbound procurement agent.

From phone calls to the inbox

The business side of the release is where Yelp's AI revenue lands. Smart Lead Reply is a paid product at $30 per month that answers incoming customer messages and qualifies the lead, with the business owner configuring how the agent should handle the conversation. Yelp also added booking integrations with ServiceTitan and Housecall Pro, and lead-management integrations with Jobber and Workiz.

Why it matters for builders

Voice agents are moving out of demo videos into default consumer flows, so the engineering pressure shifts to telephony reliability, latency and interruption handling rather than model quality. Handling background noise, hold music and a rushed host is the real product.

The agent-to-agent gap is the next integration surface. Whoever ships a sane bot-detection and escalation protocol for voice calls will sell it to every company deploying receptionist or outbound agents.

And narrow scope wins. Yelp did not ship a general assistant that calls anyone about anything; it shipped one question and one output channel. That is the pattern worth copying when an agent has to touch the real world.