[SeoulTech x Donga.com Joint Project] Seoul National University of Science and Technology (hereafter SeoulTech) operates various support programs to assist the growth of startups, including preliminary and early startup packages, maker spaces, and global collaboration. Furthermore, in partnership with Donga.com, it provides global news to aid startups in overseas promotion and expansion, introducing promising deep-tech startups domestically and internationally.

One of the challenges small and medium-sized enterprises (SMEs) face in the export process is finding buyers and local partners. SMEs spend significant time and resources searching for overseas partners. However, traditional sales methods like participating in exhibitions and networking often result in low transaction success rates and small-scale contracts. This is why AI sales agents are gaining attention in the overseas sales market. AI sales agents automate overseas customer recommendations, sales message delivery, and meeting connections, allowing SME sales personnel to secure buyers without direct effort.



Many SMEs attempt to export, but the probability of closing a deal with a buyer is low. According to data released by the Statistics Development Institute in 2024, one in three small business owners attempting to export fails within a year. One reason is the inability to find buyers willing to purchase their products.

DeepSales is a startup providing business-to-business (B2B) overseas sales agent services. Utilizing global export-related data and deep learning algorithms, it supports companies' export sales activities. This solution learns manufacturer information and predicts potential buyers with high transaction success rates to assist companies in expanding overseas. How has DeepSales brought innovation to the corporate export sales environment? An interview with Jinseong Kim, CEO of DeepSales, was conducted.

AI Sales Agent 'Selly' Navigating the Global Stage Uninterrupted

"Starting with an online marketing business during university, I gained diverse business experiences in Silicon Valley, Hong Kong, Shanghai, the US, and Europe. Through this, I deeply understood the difficulties of SMEs' sales activities. The problem lies in the overseas sales structure. Most SMEs hoping to expand overseas face difficulties from the buyer discovery stage. To solve this issue, I focused on AI and big data, leading to the establishment of DeepSales."

Jinseong Kim founded DeepSales after witnessing SMEs with competitive products and services struggle at the initial stage of overseas expansion, 'buyer discovery.' He focused on fragmented buyer information, believing that gathering scattered global information and building information reliability could secure corporate competitiveness. He emphasized the importance of transforming data into meaningful sales opportunities rather than merely providing it.



The solution DeepSales proposed to solve SMEs' export dilemma is the AI sales agent Selly. Selly is designed to function as a virtual Business Development Representative (BDR) performing the role of a company's overseas sales team member.

Selly's work process is systematic. First, it learns the client company's product and business information, then analyzes a vast proprietary database to identify the most ideal potential buyers. It then writes and sends personalized emails in the buyer's native language, processes initial responses, and selects interested potential customers. Finally, when meaningful business discussions are possible, Selly arranges meetings between the client company and buyers, completing its mission.

To enable flexible work processing, DeepSales has built a sales-related database with over 80 million records, including global B2B contact information. This data was collected from approximately 500 public data sites and commerce platforms worldwide. The collected data undergoes additional refinement, preprocessing, and labeling by the in-house data engineering team to enhance reliability.

The process of predicting transaction success probabilities beyond buyer searches is handled within a proprietary algorithm called DeepRank. DeepRank uses an approach that understands industry-specific characteristics and applies different weights accordingly. For example, when finding buyers in the healthcare sector, it assigns high weights to the possession of product certifications like FDA and CE. In the fashion sector, it evaluates brand awareness-related indicators as important. This industry-specific tailored analysis enhances initial sales accuracy.

Focus on Industry-Specific Features and Enhanced Security

What are DeepSales' concerns? Jinseong Kim mentioned the technological advancement for the expanded application of AI agent Selly across industry sectors as a current major challenge. The goal is to create an 'industry-specific expert sales agent' that deeply understands the characteristics and sales methods of each industry sector, such as cosmetics, electronics, machinery, and software, rather than a general sales AI agent.

Another challenge is maintaining high data quality and compliance with regulations in each country. To operate in the global market, capabilities to respond to differing data quality standards and privacy regulations by country are necessary. DeepSales adheres to international information protection standards such as the General Data Protection Regulation (GDPR), the California Consumer Privacy Act (CCPA), and ISO 27001/27701, but active responses to the continuously changing international regulatory environment are needed.



DeepSales has set plans to address these challenges and has begun execution. Selly is in the process of building specialized models that learn the terminology, sales cycles, and customer characteristics of each industry. The focus is on providing an experience as if the customer is working with a veteran sales expert in the respective industry.

Plans to segment Selly's roles are also under review. For example, agents that discover buyers, agents that write customized emails, and agents that manage sending and responses each perform their roles and collaborate organically like a team. Jinseong Kim stated, "We are envisioning Selly operating as an orchestration of multiple specialized agents rather than a single AI agent."

Efforts to expand language support preferred by buyers are also underway. Currently, Selly supports five languages: English, Chinese, Spanish, Japanese, and Korean, with plans to add French and Arabic. To enter the North American, European, and Middle Eastern markets, efforts to strengthen local buyer networks and partnerships are being accelerated.

Supporting Successful Global Trade Activities with Selly

DeepSales has built growth momentum by increasing its total user count threefold and its paying customer count more than fourfold over the past year. Jinseong Kim stated, "DeepSales' customers have achieved results such as securing buyer information more than twice as fast compared to traditional sales methods and increasing sales opportunity conversion rates by over 20% to 30%. We will focus on technological advancement to enable companies to achieve more sales results."

DeepSales' achievements have also extended to the investment market. Following seed investment from Bluepoint Partners, DeepSales secured a pre-A investment of KRW 2 billion from Yes24 and Woori Technology Investment. The company has also received awards and selections in support programs, including the Minister of SMEs and Startups Award, the Grand Prize in the Public Data Utilization Contest, and selection in the AWS Generative AI Spotlight Program.



SeoulTech's global support program acted as a catalyst for DeepSales' entry into overseas markets. Supported by the AWS Jungle Program, DeepSales gained opportunities to participate in global exhibitions and access local buyer networks. They directly connected with local buyers in the US, received feedback on their services, and generated sales. Jinseong Kim stated, "With the help of SeoulTech and the AWS Jungle Program, we gained confidence that DeepSales' technology and business model can succeed overseas."

"It is estimated that there are about 200,000 overseas sales personnel in Korea. The average annual salary is about KRW 50 million. Calculating this, the labor cost for overseas sales personnel amounts to about KRW 10 trillion. I anticipate that this market will gradually become agent-based. The agent market is growing by more than 40% annually. DeepSales aims to become a leading company in the sales agent field."

Jinseong Kim aims to establish a foundation where all companies can engage in efficient and easy data-driven global trade. The AI sales agent Selly is expected to open new avenues for SMEs' global expansion.

IT Donga Reporter Hyungseok Kang (redbk@itdonga.com)

Popular News