Inside Rakuten AI: Guangda Li on reinventing customer experiences with GenAI
In this series, we sit down with Rakuten AI leaders for a deep dive into the stories behind this transformative technology and the inspiring individuals driving Rakuten’s vision of AI for all. Watch this interview and others in the Inside Rakuten AI series on our YouTube channel.
Generative AI (GenAI) is revolutionizing how we shop, travel and live, offering hyper-personalized experiences that delight customers and drive growth for online merchants.
In 2025, Salesforce revealed that AI and AI agents influenced $229 billion—or 19%—of global online sales during the 2024 holiday season, a 6% increase from the previous year. From product recommendations and targeted offers to conversational customer support, AI is reshaping the digital marketplace.
Leading Rakuten’s ambition to enable AI-driven customer experiences is Guangda Li, General Manager of the AI Lab at Rakuten Asia’s AI & Data Division. As the mastermind behind the Rakuten AI Assistant, Li is redefining how AI can serve as both a trusted advisor and a powerful guide through Rakuten’s expansive ecosystem.
A vision for profitable growth: from search to AI
A veteran technologist and entrepreneur, Li has spent over a decade turning emerging technologies into profitable products. His career in the startup world saw him involved in everything from computer vision to deep learning. “Previously, I developed and sold visual search technology to enterprise users. If you look at a lot of e-commerce apps today, there will be a camera button in the search bar. When you take a photo of a product with this function it will find it or similar products,” he explained. “This how visual search works and I was involved in this tech that became very popular and is now used by many different companies to boost sales.”
“RAKUTEN’S ECOSYSTEM PROVIDES AN IMMENSE WEALTH OF DATA AND EVENTS. WHILE MANY CAN DEVELOP THIS TECH, OUR SEAMLESS INTEGRATION OF DATA WITH DIVERSE SERVICES SETS US APART FROM THE REST.”
GUANGDA LI, AI LAB GENERAL MANAGER AT RAKUTEN ASIA’S AI & DATA DIVISION
In 2023, Li joined Rakuten, drawn by the company’s ambitious vision and diverse ecosystem of services. Initially, his work focused on keyword-based search, but his relentless vision for more intuitive user experience solutions soon sparked a bigger idea: what if there was a single, unified entry point where users could effortlessly access all of Rakuten’s services and receive the most relevant information? The answer, as it turned out, lays in AI-powered chat.
“Our first demo was a simple search interface that used an algorithm to automatically determine the type of service or product to search for. From there, we began exploring ways to make the interaction more dynamic—allowing users to input free text, images, or even voice commands,” Li recalled. “That’s when it hit us: this could evolve into a dynamic chat interface—like ChatGPT, but tailored to Rakuten’s ecosystem. When we showcased the updated demo to Rakuten Group stakeholders, the response was overwhelmingly positive, giving us the confidence to press forward with our concept.”
Building AI-powered services that drive monetization
Li’s team began building the Rakuten AI Assistant with a pragmatic, federated approach, blending in-house and external technologies to deliver a high-quality product with the best possible user experience. The assistant is powered by three core technologies that not only enhance usability but also drive monetization:
- Large language models (LLMs) are used to understand user questions and convert them into actionable queries, which are then connected to Rakuten’s various search engines to deliver relevant products or information. Ensuring users find exactly what they need increases user satisfaction and retention, which are critical for driving repeat purchases.
- Semantic search technology, developed in-house, enhances the search experience by understanding the meaning behind user queries, enabling a more intuitive and context-aware interaction. It serves as the foundation for Rakuten AI Assistant’s universal concierge functionality. By understanding the intent behind user queries, semantic search delivers highly relevant results which encourage users to explore more products and services.
- Personalized recommendation feeds developed by Rakuten’s recommendation team offer users infinite scroll feeds with visually engaging product and service suggestions. This stream of tailored recommendations keeps the user engaged while creating opportunities for cross-selling and upselling.
Together, these technologies create a seamless, intuitive experience that keeps users coming back—an essential factor in increasing Rakuten’s average revenue per user (ARPU).
Leveraging Rakuten’s advantages to create a unique assistant
With a growing number of AI chat assistants available today, how does Rakuten AI Assistant set itself apart? Li highlights the service’s unique foundation: “Rakuten’s ecosystem spans a wide range of businesses, providing an immense wealth of user data and events that enable us to continuously refine the technology. While anyone can develop technology, it’s the seamless integration of this data with our diverse services that sets us apart from the rest.”
“LLMs ARE GREAT FOR GENERAL KNOWLEDGE, BUT LACK UP-TO-DATE OR NICHE EXPERTISE. WE USE RAG TO PULL ACCURATE, REAL-TIME DATA TO GIVE PRECISE, DATA-BACKED ANSWERS THAT ENCOURAGE USERS TO CONTINUE EXPLORING.”
From e-commerce on Rakuten Ichiba to second-hand shopping on Rakuma, Rakuten AI Assistant connects users to nearly every aspect of daily life. Li envisions it as a true companion, combining advanced technology with Rakuten’s ecosystem to deliver a personalized, comprehensive experience.
For example, Rakuten AI Assistant can recommend marathon shoes based on cushioning, carbon plates and price range, saving users hours of research while increasing the likelihood of a satisfying purchase. Li himself used it to plan a trip to Hokkaido, receiving hotel suggestions, travel tips and product recommendations—all in one place. This convenience encourages users to rely on Rakuten for multiple aspects of their lives, increasing engagement across services.
Building trust with a habit-forming sticky AI product
Creating a powerful AI assistant is only half the battle; Li’s team also focuses on building trust and making the assistant a habit-forming tool. As a sticky product, Rakuten AI Assistant has the potential to continuously unlock further revenue growth.
“WE’RE EXPLORING HOW TO INTEGRATE MORE RAKUTEN SERVICES AND MOVE TOWARD AI-NATIVE USER EXPERIENCES.”
“Large language models are great for general knowledge, but they lack up-to-date or niche-specific expertise,” Li notes. “That’s why we use Retrieval-Augmented Generation (RAG) to pull accurate, real-time data from internal and external sources. When users get precise, data-backed answers, they’re more likely to follow up with more questions and continue their exploration process.”
Li is also locked in on the idea of super personalization. “Many AI assistants today have short-term memory focusing on immediate transactions, but we’re developing our assistant’s long-term memory,” Li said. “Rakuten AI Assistant should recognize you and understand all your purchase history, behavior and interactions to anticipate your next wish based on your unique preferences. It’s like having your own genie in a lamp, always ready to deliver exactly what you need. To me, this level of super personalization is the holy grail of AI assistant services.”
The future of AI-native experiences
Li’s ultimate vision is for the Rakuten AI Assistant to become an indispensable, AI-native experience. By recognizing users through their Rakuten IDs, the assistant can seamlessly connect them to the ecosystem, offering timely, relevant recommendations.
“We’re exploring ways to systematically integrate more Rakuten services and move toward AI-native user experiences,” Li shares. “Imagine opening your phone and being greeted by an assistant that knows exactly what you need—whether it’s a product recommendation, travel tip, or service update. Interactions should feel natural and friendly, not forced or robotic.”
Going forward, Li is eager to leverage a unique blend of innovative technology, deep personalization, and seamless integration, positioning Rakuten AI Assistant as a value-creating service that redefines how users interact with AI—and with Rakuten itself.
