Retail spent two years bolting chatbots onto websites that could field a shopper’s specific questions but never close a sale. Voice agents built on models like OpenAI’s GPT-Realtime are changing that, holding real sales conversations, asking the same probing questions a floor associate would and adapting as a shopper’s needs shift, all without a human ever stepping in.
Japanese home-appliance retailer Yamada Denki turned its associates’ sales expertise into a 24/7 multilingual voice agent built on OpenAI’s GPT-Realtime, OpenAI reported. During a two-week public campaign on Yamada Denki’s online store, about 30,000 people used the agent, and 92% of post-conversation survey responses were positive.
“This is not an extension of the conventional chatbot,” said Akira Fukabori, CEO of avatarin, the artificial intelligence customer service company that developed the voice agent. “It is an attempt to expand what customer service can be with AI. We see it as the beginning of an ‘interface revolution’ in retail,” he added.
Voice Agent Asks Questions Instead of Just Answering
The distinction that separates this from a chatbot is structural. A conventional chatbot waits for a shopper to type a specific keyword before it can respond usefully. avatarin’s agent listens for context instead, asking follow-up questions to uncover what a customer actually needs, the way a trained associate would.
“Customers do not want a chatbot. They want intelligence,” Fukabori said. “‘I need a refrigerator for a family of four, but my kitchen is small. Which one should I choose?’ Real customer service means being able to answer that question.”
A retrieval-augmented generation system grounds the agent’s answers in real product data. GPT-Realtime handles voice, text and image understanding within a single low-latency conversation, letting the agent stay responsive even as a shopper changes requirements mid-conversation, according to OpenAI’s description of the process.
In practice, customers described the agent as “easier to talk to than an actual sales associate” and said they appreciated being able to ask questions again without feeling stressed, OpenAI reported. The format also let shoppers speak candidly about budgets or uncertainty without feeling sales pressure, particularly after stores had closed for the day.
Consumers Are Already Comfortable Letting AI Close the Sale
The shift lines up with where consumer behavior already sits. Nearly half of online shoppers, 48%, used AI to research their most recent purchase, according to PYMNTS Intelligence research conducted with Visa Acceptance Solutions across the U.S., Brazil and the UAE.
That trust extends beyond research: 45% of consumers say they would be comfortable letting an AI agent complete a purchase on their behalf, a figure that rises to 54% among Gen Z, PYMNTS Intelligence found separately, with 81% of retailers saying they already trust AI to operate autonomously once the right guardrails are in place.
Retailers are responding to that shift with real budget commitments: 43% of retailers are actively piloting autonomous AI shopping agents, PYMNTS Intelligence reported.
The Yamada Denki case gives that investment a concrete result to point to: an agent that asked the right questions, in a shopper’s own language, at any hour a store was closed, and left the majority of people who used it satisfied enough to say so.
For all PYMNTS AI coverage, subscribe to the daily AI Newsletter.