ИИ-агент avatarin на GPT-Realtime обслужил 30 000 клиентов Yamada Denki

Japanese startup avatarin has deployed a retail AI agent based on OpenAI's GPT-Realtime API in stores of major electronics chain Yamada Denki. In the first two weeks of operation, 30,000 shoppers used the agent, and 92% of survey respondents rated the interaction positively — such data is published by OpenAI Blog in a case study about the project.

What the agent does in Yamada Denki stores

The avatarin AI agent advises shoppers 24/7 without the involvement of live staff. The system takes on the role of a consultant during nighttime hours, on holidays, and during peak load periods when there are not enough employees. At its core is OpenAI's GPT-Realtime API: an interface designed for real-time conversations with minimal latency and support for voice and text input.

The agent's key advantage is multilingualism. For Japanese retail, this is critically important: Japan receives tens of millions of foreign tourists from Asia, Europe, and North America every year, and the language barrier has traditionally been a significant limitation in service. The agent allows shoppers to address it in their native language without involving a translator or a specially trained employee.

"Avatarin uses

OpenAI's GPT-Realtime to give Yamada Denki customers round-the-clock multilingual support," states the official OpenAI Blog publication about this project.

Key figures for the first two weeks

The avatarin agent reached 30,000 users in its first two weeks of operation at the Yamada Denki chain. According to OpenAI Blog, the results are as follows:

• 30,000 shoppers used the agent in the first two weeks
• 92% of survey participants rated the interaction positively
• Operating mode: around the clock, without breaks or days off
• Multilingual support without hiring additional staff

A 92% approval rate is a high result for any automated customer service channel. Traditional chatbots in retail often receive low ratings due to templated responses and limited understanding of the context of a request. GPT-Realtime API, according to OpenAI, enables more flexible dialogue: the system understands the nuances of a question and responds according to the situation.

Why GPT-Realtime and not an ordinary chatbot

OpenAI's GPT-Realtime API was created specifically for low-latency scenarios — voice agents, live consultations, interactive real-time assistants. Unlike the standard pipeline, in which speech is first converted to text, then processed by a language model and voiced again, GPT-Realtime processes the audio stream directly. This reduces latency and makes the dialogue more natural for the user.

For avatarin, choosing this particular API allowed the architecture to be simplified and the launch to be accelerated. According to the OpenAI Blog publication, the team implemented an industrial solution in a short timeframe — without the long corporate deployment cycle typical of traditional customer service systems. This approach lowers the technical threshold for companies that want to add a voice AI agent but are not ready to build their own speech recognition pipeline from scratch.

What this means

The avatarin and Yamada Denki case is one of the first public confirmations that the GPT-Realtime API works under conditions of high traffic in real retail with measurable results. 30,000 interactions in two weeks and 92% positive ratings — this is not a laboratory pilot, but a production result in an operating store network.

Retail chains facing staff shortages, the need for round-the-clock coverage, and language barriers receive a concrete example of how the GPT-Realtime API solves all three challenges simultaneously. avatarin has shown: the path from idea to an industrial AI agent with high user satisfaction has become significantly shorter.