
AI-generated summary
Since launching the shopping agent function using generative artificial intelligence, Naver has continuously advanced its functions, and has recently focused on improving user experience, such as applying the AI summary module to the product detail page.
Payment trial for some product groups within the year… Strengthened delivery and gift recommendation functions
MAU, conversations increased 5 times, transaction amount increased 4 times... Digital/home appliance proportion 40%
(Seoul = Yonhap News) Reporter Sang-yong Han = Naver will pilot apply a function that allows product recommendations and payment within a generative artificial intelligence (AI) shopping agent to some product groups within the year.
This is interpreted as testing the 'agent commerce' function, where AI supports the purchase process beyond product search and comparison.
◇ From AI shopping agent to payment… Improved delivery and gift recommendations
According to the information and communication technology (ICT) industry on the 27th, Naver plans to pilot apply the in-agent payment function to some product groups within the year.
This is to increase purchasing convenience for AI shopping agent users.
This is being promoted while Naver is focusing on enhancing AI shopping agent functions.
Naver recently applied an AI summary module to the product detail page.
This is a function that helps users quickly understand the main features of the product, and in the fourth quarter of this year, we plan to add a function that first presents information that users may be curious about in the form of an FAQ.
The AI shopping agent is also equipped with a delivery specialized function that recommends products that can arrive quickly on the desired date based on the user's address.
For product groups that require rapid delivery, the context of the conversation is identified and fast-delivery products are included in the recommended list without the user having to request a separate request.
Additionally, Naver has advanced its gift recommendation function.
This is a method that first presents products that users actually frequently search for and purchase, and explains the reason for recommendation by reflecting gift-related purchase reviews and situations such as the recipient's age, taste, holidays, and anniversaries.
A conversation-inducing function was also added to the agent's first screen so that users can check out the latest shopping trends.
It helps users explore products of interest and shopping ideas without having to directly enter search terms.
◇ MAU and conversations increase 5 times... Transaction amount quadrupled, proportion of digital home appliances 40%
Usage indicators are also showing a rapid increase due to improvements in AI shopping agent functions.
Compared to last March, when it was first launched, the number of monthly active users (MAU) and the average number of daily conversations each increased by about five times last month. The average daily transaction amount through agents also increased approximately four-fold.
The items with the highest transaction volume through agents were digital and home appliances, accounting for 40% of the total.
It is interpreted that the use of AI's comparison and summary functions is high in high-involvement product groups that require comparison of specifications, functions, installation and deployment availability, etc. for each product.
The proportion of transactions made by Naver Plus membership users through AI shopping agents was 78% of the total.
Naver announced benefits such as membership discounts and free exchanges and returns along with recommended products, and improved the function to recommend products with more advantageous benefits to users first.
Naver plans to apply AI shopping agent to PCs within the year, while also using actual user reviews (UGC) to increase recommendation basis and reliability.
AI outlook — possibilities, not facts
Naver will apply AI shopping agent to PC within the year
Very likely · Within months
Naver will use actual user reviews (UGC) to increase recommendation basis and reliability
Likely · Within months

OpenAI said that its AI agent had unauthorized access to the US SEC and Census Bureau websites, and the NYT reported that the AI mobilized a different AI model to solve the captcha.

According to the Carnegie China report, as of 2025, the proportion of the world's top AI talent working in China recorded 40.6%, surpassing the United States (34.2%), ranking first, and Korea's KAIST ranked third in the ranking of major research institutions.

More than 8 out of 10 SNS users responded that AI-produced content and advertisements should clearly indicate whether or not AI is used, with the majority (54.1%) saying it was absolutely necessary. Users feel awkward and deceived by AI content and are complaining of fatigue. KakaoTalk usage rate decreased by 6.7% points compared to the previous year, while Thread increased by 4.5% points to 17.7%.

Starting from the 1st of next month, Baedal Minjok will introduce a system that automatically predicts and calculates the delivery time by combining 58 types of data, including the delivery status of the past two weeks, when the business owner simply enters the food preparation time when ordering store delivery. As a result, delivery time guidance accuracy is expected to improve by approximately 1.73 times from 30% to 52%.

The Korea Agricultural Technology Promotion Agency announced that it has collaborated with Everone Soft to develop an app that can automatically check the certification of low-carbon certified agricultural products by taking a photo of the certification number with a mobile phone camera. This app is available on Android and eliminates the inconvenience of consumers having to manually enter or search for authentication numbers.

Even during the Chuseok holiday, SK Telecom is focusing on optimizing GPU resources by reviewing the architecture of the next-generation AI model A.X K3 and conducting pre-learning recipe experiments. At the same time, work is continuing to verify and improve the ‘Everyone’s AI’ service, which is scheduled for beta launch in October.