Lotte Homeshopping case study

Lotte Homeshopping is a household name in South Korea for good reason. Part of the Lotte Group, a top-5 conglomerate spanning retail, food, hospitality, construction, finance, and IT, Lotte touches nearly every aspect of daily life in Korea. With over 80 affiliated companies and operations in more than 30 countries, the group’s scale rivals global giants like Walmart or Amazon.

As the group’s flagship for televised and digital retail, Lotte Homeshopping curates high-quality, trend-driven products delivered to millions of customers via TV broadcasts, mobile apps, and online platforms. Behind this seamless customer experience lies a meticulous quality assurance (QA) process, essential for verifying product details, ensuring compliance, and meeting tight broadcast schedules. However, as volumes increased—particularly within their fast-paced fashion division—the QA process began encountering significant friction, slowing down product launches.

Retail’s challenge: Shorten lead times without sacrificing quality

Lotte Homeshopping's strength lies in quickly sourcing trendy products and delivering them to customers through engaging broadcasts. QA is an essential part of the broadcast production process, but there were several friction points:

• Manual communication slowed product approvals across phone, email, and messenger.
  

• Delayed responses to partner questions dragged down workflows.
  

• Overworked QA agents were bogged down by repetitive inquiries.

These challenges were most apparent in the fashion division, with its seasonal product changes, new trends, and multiple partners.

To move faster, Lotte needed a solution that could automate partner support, verify documentation, and scale without compromising on accuracy or compliance.

Accurate, enterprise-ready agentic AI

Lotte Homeshopping built an AI agent named Moni using Sendbird’s AI Agent platform, powered by Anthropic’s Claude 3.5 Sonnet LLM. After testing multiple vendors, Sendbird stood out for its enterprise-readiness and developer-friendly design.

Standout Sendbird + Claude features:

• Reliable, accurate QA support: Anthropic’s Claude is a trustworthy LLM, and Sendbird’s platform enables fast AI agent development and deployment. Together, they delivered a high level of accuracy and transparency essential to QA work, such as reviewing product test reports, labeling, and web documentation.

• Streamlined partner collaboration: QA agents previously spent a lot of time contacting multiple partners to verify product descriptions or obtain missing or incomplete information. Using Moni, suppliers can easily access and provide supplementary data as required, eliminating the time and effort associated with manual follow-up.

• Scalability: Anthropic’s clear documentation and release updates make it easy to deploy new features and scale quickly. The platform provides a strong foundation for Lotte Homeshopping to scale future AI services.

• Analytics and observability: Sendbird’s dashboard and analytics tools provided Lotte Homeshopping’s AI Promotion Team with the necessary information to build and fine-tune the Moni AI agent.

• Enterprise-ready security out of the box: Sendbird stood out as the only AI solution in its class with built-in enterprise-grade security, including features like message moderation and two-factor authentication (2FA). For a large enterprise like Lotte Homeshopping, these capabilities were critical in meeting internal compliance standards and accelerating deployment.

The Sendbird AI Agent platform promised efficiency, compliance, and accessibility with quick deployment. Meanwhile, testing revealed that Claude responded effectively and accurately to the administrator’s unique needs. It was a one-two punch that would supercharge their customer experience while laying a trusted foundation for using various AI in retail going forward.

Implementing Sendbird to ease QA headaches

Lotte Homeshopping began this journey by spending six months testing various solutions. When the company discovered Sendbird AI, the project gained momentum rapidly. From that point, the retailer took only a few months to complete a proof of concept (POC), implement the solution, and onboard its fashion partners. Compared to other product groups, fashion has a 24/7 QA process and guidelines for responding, a limited number of retrospective points compared to other product lines, clear learning standards, and the largest number of partners, so it was an easy choice to start with fashion partners first.

The combination of Sendbird’s friendly AI agent UI, simple embed code, and Anthropic Claude’s reliable LLM integration made deploying the solution quick and easy. The streamlined process gave the company’s AI engineers more time to focus on creating prompts and establishing the baseline. The team trained Moni on standardized knowledge across all product categories. This training enabled Moni to provide consistent answers to any inquiry, improving compliance management.

Results: More efficient QA, happier partners

Just four months after introducing its Moni AI agent, Lotte Homeshopping’s QA process has undergone a remarkable improvement. All inquiries from new and existing partners are handled through Moni, which preemptively checks for issues before they arise, significantly improving operational efficiency.

• Reduced inquiries directed to human agents: On average, Moni answers 53 inquiries daily, reducing the need for human responses by 30–40%.

• Gained 24/7 support: Partners from across the globe can direct questions to Moni and receive immediate, accurate responses.

• Faster time-to-market: A survey of fashion partners revealed that since Moni’s introduction, delays from inception to launch have decreased dramatically.

• Improved partner and employee satisfaction: Partners love the increased efficiency. Feedback from internal evaluations was so positive that the company is now expanding Moni to other products.

The Sendbird AI agent is more than just a QA chatbot. It supports internal chatbots, such as those used for corporate IT support, and it also serves as a test bed for potential customer service and call center applications.

What’s next: Scaling AI across the business

Moni’s roadmap includes fully automated test result verification, expanded use in marketing and broadcast production, and even training human QA agents for improved consistency.

Sendbird & Lotte: Raising the standard for enterprise AI

Lotte Homeshopping is continuously exploring new ways to elevate its customer experience (CX). The company actively evaluates various LLMs and AI models, both locally and across cloud platforms such as Amazon Web Services (AWS), to identify new business applications. Sendbird’s omnichannel AI Agent evolves alongside these efforts, regularly adding features that help Lotte Homeshopping enhance CX and quickly scale new opportunities. With its proven scalability, straightforward integration, and robust security standards, Sendbird enables Lotte Homeshopping to deliver the exceptional service its customers expect.