On the AI Frontlines: How Rakuten and industry leading startups are shaping the future of tech
AI generated movies. Automated customer service agents. Personalized shopping assistants. These are just a few ways artificial intelligence is seamlessly integrating into everyday life.
But beyond the excitement, global AI innovators must come together to address crucial questions: What are the implications of rapidly evolving AI technologies and where will they lead us? How can we shape these technologies to bring true value to society? How can we confront real issues around ethics, data privacy and the significant costs of a single query?
At Rakuten AI Optimism 2025, Rakuten’s Chief AI & Data Officer Ting Cai joined Silicon Valley innovators Jesse Zhang, Co-founder and CEO of Decagon, and Tom Graham, Founder of Metaphysic, to discuss the current state of AI, challenges and trade-offs of AI models, and how the technology will continue to impact industries of the future.
AI is changing our daily lives – from entertainment to customer service
The panel kicked off with a movie trailer for “Here”, a $50 million production that used real-time generative AI to seamlessly de-age stars Tom Hanks and Robin Wright. The visual effects were powered by Metaphysic’s advanced AI technology.
Graham noted how rapid advancements in fine-tuning and data science are producing photorealistic images that are increasingly difficult for audiences to distinguish from AI, CGI or traditional VFX (visual effects). At its current pace of development, AI-generated content is expected to transform not just movies, but also TV, music and social media in the coming years.
“In the movie context, this is going to dominate how we create content going forward,” said Graham, “It will become much cheaper and, in fact, better.”
But it’s not just entertainment. Customer service is also adopting powerful AI solutions and attracting investor interest.
Decagon, a leading startup building customer support chatbots and agents, automates tasks such as canceling subscriptions, processing refunds and answering FAQs. Its AI-powered bots have delivered measurable results for clients such as ClassPass, which reported a 95% drop in cost per reservation after implementation.
“For people that are interacting with banks, airlines and grocery stores, our AI agents are there to answer and solve your issues,” says Zhang. AI chatbots are becoming more advanced and human-like as developers integrate more feedback and fine-tune their models.
A shift toward specialized AI models
Early AI models handled a single task, but large language models can now manage many at once. This raises the question of whether one universal model will prevail or if the future lies in specialization.
The panelists agreed that as organizations look to scale, they are turning to specialized models to balance speed, cost, efficiency and accuracy. Cai highlighted that meeting user expectations like near-instant search results, often demands this careful trade-off.
In customer service, Zhang stressed that reliability matches performance in importance. He advocated combining general models for early versions, with specialized ones for specific needs, supported by rigorous testing, real-time monitoring and post-interaction review.
Rakuten AI: Balancing cost, speed and quality
At Rakuten AI Optimism, Group Chairman and CEO Mickey Mikitani made waves when he announced the full-scale launch of Rakuten AI, a new AI agent available for free on the web (beta) and in the Rakuten Link app for Rakuten Mobile subscribers, with plans to expand across the Rakuten ecosystem, including Rakuten Ichiba, this fall.
During the panel, Cai highlighted that Rakuten AI is designed to balance cost, speed and accuracy, with a deep specialization in Japanese language and culture. He explained that the model’s rigorous multi-stage training and ongoing human feedback ensures it adapts quickly to user needs.
By leveraging high-quality Japanese data and focusing on efficiency, Rakuten AI delivers reliable performance tailored to the Japanese market.
“We want to build the most cost-efficient model while keeping performance high, so we can pass the most benefit to our customers, whether individual consumers or enterprise,” Cai said.
As industry leaders like Rakuten, Decagon and Metaphysic continue to push boundaries, the focus remains on balancing cost, speed and quality, while keeping user trust at the forefront. The future of AI will depend not only on technological breakthroughs, but also on collaboration and thoughtful stewardship as the world moves forward into a new era of intelligent solutions.
