Rakuten AI: Our agentic future starts here

At the 2025 Rakuten AI Optimism conference, Rakuten took another bold step into the future with the full-scale launch of Rakuten AI, a new cutting-edge AI agent geared to reshape how users interact with the Rakuten Ecosystem.

“We are on a journey of AI innovation,” shared Rakuten Group Chief Data & AI Officer Ting Cai presenting on day one of the flagship event. “We are infusing AI into everything we do. We believe that this has more potential to benefit all of society than any revolution before it.”

The next evolution of Rakuten AI

The full-scale launch of Rakuten AI represents more than a technical upgrade; it’s a movement to centralize AI capabilities across Rakuten’s diverse portfolio of services. The new platform goes beyond the functions of a typical AI chatbot to serve as a seamless, intelligent entry point to the Rakuten Ecosystem, spanning sectors like shopping, lifestyle, travel, fintech, education, wellness, entertainment and more.

In addition to assisting with research or shopping, Rakuten AI is built to understand local context through deep Japanese language and cultural knowledge, aiming to deliver highly personalized, localized and relevant support. The new AI agent also integrates features like natural language chat, voice-to-text input, image-based queries, and smart prompt suggestions to streamline the user experience.

Initially available via the Rakuten Link mobile app for Rakuten Mobile subscribers, Rakuten AI is also accessible through a standalone web app (beta) for anyone in Japan, free of charge. Later in 2025, the agentic AI tool is set to roll out on Rakuten Ichiba, Japan’s largest online marketplace.

During his keynote, Cai welcomed members of Rakuten’s AI product team, represented by Caitlin Cai and Cade Reid, to give a live demonstration. The demo showcased Rakuten AI’s versatility by performing complex tasks such as conducting market research for a soy sauce product, planning the launch of an online store on Rakuten Ichiba, and even generating cooking recipes using the product.

Cai and Reid demonstrated a few particularly impressive Rakuten AI highlights, such as the problem solver feature which can analyze an image of a mechanical engineering math problem and provide a detailed solution with step-by-step explanations.

Rakuten’s Agentic AI era

Underpinning Rakuten’s vision is the idea of an agentic ecosystem. Rakuten AI no longer only answers questions and gives recommendations; it seeks to understand your intent, drawing on trillions of interactions across Rakuten’s services, and taking action.

“Some people say the future is agentic. Agents are everywhere. But what does that mean?” Cai posed. “I personally feel that AI is growing its capability. From being reactive, answering questions, using it to find certain facts – now it can proactively act on your behalf.”

Cai highlighted three major shifts that define agentic AI: “First is moving from chat to task. Second, from research to action. And third, from human browsing to agent browsing.”

This shift is already being reflected in how platforms are built. “For over 30 years, we designed sites to be operated by humans. Now, agents are able to browse the web and take actions on your behalf.”

The new Rakuten AI fully embraces this agentic evolution.

“Whether it’s booking hotels or purchasing items, agents can better understand your intent and get things done for you,” Cai said. “Last year at Optimism, we announced that Rakuten AI could already take actions. And today we’re putting that capability into more products.”

The new Rakuten AI offers a suite of capabilities: deep-think search, coding support, translation, image generation, and more – not simply features tacked onto an existing model, but integrated abilities intended to provide seamless utility.

“We built it with deep learning, tokenization, large language models – but most importantly, products. Products that solve real problems and leverage our channels to reach billions of customers,” Cai said. “Rakuten AI is a powerful agent for your personal and professional life. It offers the full power of generative AI and a seamless integration experience with the rich Rakuten Ecosystem.”

The long-term goal is to democratize this power. “With AI, now everyone can become a software engineer, a business owner, or open a shop on Rakuten Ichiba. So we envision a future where Rakuten AI can become the intelligent hub to connect all of the Rakuten Ecosystem services to bring more value to you.”

Retooling the very foundations of the Rakuten Ecosystem

Delivering on this agentic vision required a foundational approach. Rather than bolting AI onto legacy systems, Rakuten has been retooling its tech stack from the ground up.

“We have done so much at Rakuten. This includes investment in fundamental technologies – like for engineers and product managers, understanding the fundamental changes in the technology revolution, so we know what is possible and what is not.”

Cai stressed that a key part of the new platform is Rakuten’s custom language modeling work, which centers on Japanese language and culture. “We built a tokenizer for Japanese specifically so we can capture the language in a very fundamental way.”

From that base, Rakuten developed its own deep learning architecture and built embeddings for every entity in the ecosystem. “Embeddings are important because they capture the essence of what the user wants and what the product represents.”

This foundational work is already paying dividends. On Rakuten Ichiba, the introduction of AI-powered semantic search drove a 10.7% year-on-year boost in gross merchandise sales attributed to search.

Rakuten Ichiba is also looking to implement personalized search, which optimizes results based on individual users’ preferences. “Now we can capture preferences and tastes based on past searches and brand preferences, and recommend the right product for you.”

Product recommendations have also been given the semantic treatment, Cai revealed. “We’ve seen that these recommendations help users discover products they love, but might not have known about. On Rakuten Ichiba and Rakuten Rakuma, we’ve seen that they are purchasing more, that it benefits our merchants, delights our customers, and brings more value to everyone.”

Rakuten is busy applying these learnings to another major pillar of the Rakuten Ecosystem: advertising. “We’re now able to show more relevant ads to our customers. Even ads can become a delight, because they now match user intent.”

Sharing internal expertise

“At Rakuten, we apply everything we do. We believe that in order to build the best products, we must first try them with our own hands. From the CEO and executives to employees across the company, we are constantly using our products, testing them every day,” said Cai. “We know what’s working, what the limitations are, so we can build the best for our customers.”

Over 15,000 employees are using Rakuten AI for Rakutenians daily to write code, generate documents, translate content and more, and employees have built more than 20,000 custom AI tools and templates to date. This has transformed development speed.

“I’m most proud that our Rakuten AI for Rakutenians engineering team is using Rakuten AI for their daily jobs,” Cai said. “Before we were only doing monthly releases. Now, with our AI Assistant, we can gather requirements faster, write specs faster, debug code faster, and maintain code more efficiently. Now we’re doing weekly releases.”

The results of this embrace of AI are being extended to business customers. Rakuten AI for Business is already being used by clients like Wolt Japan and Hotel Urashima. At just 1,100 yen per user per month (including tax), it offers a compelling mix of affordability and utility.

“Our goal is to help various businesses, big or small, with their operations, whether in sales, marketing, or engineering,” Cai said. “By deeply understanding the customer’s pain points, we can build the right products to help businesses and customers have better lives.”

Merchants on Rakuten Ichiba, meanwhile, are already using the AI-powered RMS AI Assistant suite to create product descriptions, design marketing visuals, respond to customer reviews and analyze shop data.

“We also leverage our learnings to help our merchants and partners, so they can streamline their operations, bring more value to their customers, and more profit for their businesses,” Cai said. “I love these tools because they level the playing field. Before, only big companies had the resources to do this. Now, with AI, everybody can.”

One ecosystem, many agents

“Rakuten is a business, of course, but our purpose is always higher,” said Cai. “By seeing things around the corner, understanding technology fundamentals, and building awesome products we can deliver the most value for society.”

Cai concluded his address by inviting members of the conference to work together on this mission of empowerment.

“Over the last two years, we’ve been committing to building a world where AI truly augments human creativity. I think we’ve made a lot of progress. I’m super proud of what the team has accomplished,” Cai said. “AI agents are new, but the speed of execution from the team – with your help, with your feedback – is truly amazing. We hope to build an agentic AI future together with you.”