"I will recommend outfits A, B, and C based on your preferences."
At the Sendbird booth during the AWS Summit Seoul 2026 held on May 21 at COEX, an AI stylist immediately began suggesting products as soon as attendees accessed the fashion shopping mall-themed live demo screen. The AI proposed various styles based on customer preferences and previous purchase history, explaining which situations each outfit would suit. When a reporter inputted, "I want to dress more casually," the AI adjusted its recommendations accordingly.
Sendbird is a global AI communication company that provides enterprise chat and consultation solutions. Recently, it has focused on developing 'AI agent' technology that goes beyond simple chatbots to remember customer context over the long term and perform actual tasks. At this AWS Summit, the company showcased scenarios where AI handles reservations, exchanges, rebookings, and connects consultations in hotel, airline, and e-commerce environments.
One of the standout features was the 'memory-enabled AI.' All conversations and actions with customers are stored in a memory format, allowing seamless continuation even when switching channels. If a shopping inquiry starts in a messenger and transitions to a voice call, the AI retains the previous conversation details, eliminating the need for customers to repeat themselves.
During the live demonstration, the AI processed a product exchange request. It recognized the order history and previous inquiries, identified the reason for the exchange, and then connected the customer via voice call to provide pickup schedules and estimated delivery times. If complex issues or refund requests arose, the AI would connect the customer to a human representative. However, unlike traditional customer service, there is no need to explain the situation from the beginning, as the AI summarizes the entire conversation for the representative.
This functionality leverages AWS cloud technology. Voice connections and representative integration are operated on AWS Connect, designed to ensure a smooth transition between AI and human representatives within a single workflow. The core of this technology is the 'Cross-Channel AI Memory,' which maintains customer context across messaging and voice channels.
Similar processes were demonstrated in hotel and airline scenarios. In the hotel demo, the AI recommended rooms and restaurants based on the customer's stay history, family composition, and preferences, even handling reservations. In the airline demo, when a flight delay occurred, the AI proactively suggested alternative flights and managed seat reassignments and compensation notifications.
Sendbird believes that AI is evolving beyond simple response-based chatbots to perform actual operational tasks. Lee Sang-hee, CEO of Sendbird Korea, stated during a keynote session the previous day, "The future competition in AI will not be about who creates the smartest AI, but rather who can continuously remember and connect customer relationships and contexts."
Real-world implementations are increasing. According to Sendbird, Hanssem has integrated AI customer support into its order, delivery, and installation inquiries, automating over 90% of customer inquiries. Coupang Eats is also operating an AI-based customer support system.
Sendbird plans to expand its AI communication infrastructure to enable companies to build their own brand-based AI concierges.
* This article has been translated by AI.
Copyright ⓒ Aju Press All rights reserved.