What happens when conversational AI is designed not just for efficiency, but for emotion, culture, and trust?
At Convo AI World Japan, held on November 5, 2025, co-hosted by Agora and V-Cube, global tech leaders, founders, and investors came together to explore how conversational AI, avatars, and multimodal intelligence are evolving, and why Japan’s perspective matters far beyond its borders.
From real-time translation and streaming avatars to robotics heritage and cultural storytelling, the event revealed how regional strengths across Asia are converging to influence the next major leap in AI experiences worldwide. Rather than chasing scale alone, the conversations in Japan centered on meaning: how AI listens, responds, and connects with people in ways that feel natural, respectful, and human.
Here’s a summary of each session and what our amazing speakers covered in their presentations:
Session 1: Opening Remarks - Redefining Japan’s Role in the AI Era
Tony Wang, Co-founder & CRO, Agora
Tony Wang opened the event by reframing a common narrative in global AI discussions. Japan, he argued, does not need to out-scale the US or China to lead. Instead, it can redefine what leadership in AI looks like.
Japan’s strengths lie in emotion, craftsmanship, trust, and storytelling. Its cultural heritage; spanning anime, manga, gaming, music, and design, has long shaped global imagination. These strengths position Japan uniquely to build empathetic, multimodal AI that listens, adapts, and respects human nuance, rather than simply optimizing for output.
“We’re entering an era where iteration beats perfection, curiosity replaces caution, and meaning matters more than machinery, because the future won’t be won by product-market fit, but by emotion-market fit.”
Session 2: AWS AI Solutions and Real-World Use Cases
Mantaro Yamada, Solutions Architect, AWS
Mantaro Yamada explored how AI adoption is rapidly moving from experimentation to revenue-driven, production-scale deployment, particularly across media, entertainment, gaming, and digital content.
At the center of this shift is Amazon Bedrock, which enables organizations to access multiple foundation models through a single API. This unified approach reduces integration complexity and allows both technical and non-technical teams to ship AI features faster, without heavy infrastructure or model-management overhead.
He shared real-world applications, including:
- English learning platforms generating personalized conversation topics and lesson flows
- Media companies producing video highlights, summaries, and multilingual content at scale
- Gaming studios and avatar platforms maintaining consistent character personality through structured prompting
- Creative teams using generative AI for 3D assets, environment design, and rapid prototyping
“AI isn’t just boosting developer productivity; it’s transforming how writers, designers, producers, and marketers work.”
Session 3: Stepping into the Era of Streaming Avatars
Alicia Tseng, Head of Product, Akool
Alicia Tseng addressed one of the biggest challenges in conversational AI today: latency. Traditional interactions rely on slow STT → LLM → TTS pipelines, resulting in awkward pauses and robotic flow.
Akool solved this by building an ultra-low-latency streaming avatar engine that enables conversations timed to human rhythm. Today, Akool supports millions of users globally and is evolving into a full-stack platform spanning image, video, audio, and live avatars, offering hundreds of avatars, 150+ languages, and massive concurrency at scale.
She showcased real deployments across industries:
- Event guides and holograms at global conferences
- Virtual leasing agents in real estate
- Insurance avatars enabling sensitive conversations
- In-store assistants across telecom retail locations
- Healthcare avatars monitoring early symptoms
- Airline avatars assisting travelers in airports
- CEO avatars interacting directly with thousands of employees
“AI’s future isn’t about replacing humanity; it’s about amplifying it. Together, Akool and Agora are transforming chatbots into expressive, emotionally rich, real-time experiences.”
Session 4: Panel Discussion - The Future of Conversational AI and Avatars
Moderated by Patrick Ferriter, the panel brought together leaders working across avatars, infrastructure, and multimodal AI to explore what’s coming next.
Emotional Design over Pure Realism
Jia Shen, Co-founder & CEO, AKA Virtual, explained why stylized, character-driven avatars often outperform hyper-realistic ones in Japan. Users feel less judged and more open, making these designs especially effective in therapy, education, and entertainment.
Infrastructure for Real-Time AI
Yongle Yang, Head of Engineering, Dify.ai, discussed building conversational systems using fast workflows and semi-autonomous agentic models. With real-time infrastructure, developers can deploy responsive AI interfaces powered by proprietary data.
Faster, Smarter Multimodal AI
Zeyi Cheng, CTO, Wavespeed.ai, highlighted advances in near real-time diffusion models, enabling automated ad creation, conversational video avatars, and interactive content without heavy production effort.
What’s Coming Next
Panelists pointed to several near-term shifts:
- Affordable, widely adopted avatars
- AI-powered robots blending physical and conversational intelligence
- AI companions for care, connection, and daily support
- A new generation of creators building with AI-native tools
Across perspectives, one theme stood out: conversational AI is becoming expressive, emotional, and deeply interactive.
Session 5: Redefining Emotional Connection in Japanese Voice AI
Jason Chen, Head of Marketing, Jarvis
Jason Chen explored one of the most difficult challenges in conversational AI: making Japanese voice interactions feel natural over long conversations. Japanese language demands precision in pitch, politeness, context, and interruption handling, areas where many voice systems fall apart.
Jarvis addresses this through a unified platform that delivers emotional expression, accurate pitch, memory, interruptible turn-taking, and noise robustness, without heavy manual tuning. Its AI sustains personality across long interactions and feels conversational rather than scripted.
From anime IPs to tourism and enterprise deployments, Jarvis is turning voice AI from content into companion.
“Japan needs conversational-grade voice AI, and Jarvis is redefining it by bringing voice, reasoning, and memory together into one human-centered experience.”
Session 6: Tripo - Instant 3D Creation for Everyone
Frank Zhang, Tripo Japan Distributor
Tripo demonstrated how 3D creation is finally catching up with the speed of 2D AI. By turning text or a single image into clean, production-ready 3D models in seconds, Tripo eliminates long-standing bottlenecks in geometry quality, topology, and asset cleanup.
Supporting major tools like Blender, Maya, Unreal, and Unity, Tripo enables creators to generate, edit, and export assets directly into existing pipelines. A standout moment included reconstructing a broken Gundam toy part, from scan to print in minutes.
By boosting productivity and reducing costs, Tripo is making high-quality 3D creation accessible to studios, brands, and individual creators alike.
“The next leap in AI isn’t just speed; it’s giving creators the power to turn imagination into 3D reality instantly.”
Looking Ahead
Convo AI World Japan underscored a powerful shift in conversational AI. The next generation of experiences will be defined not by intelligence alone, but by emotion, culture, and real-time responsiveness. As AI becomes more embedded in daily life, the ability to communicate naturally across voice, video, avatars, and environments will determine what truly resonates with users.
Japan offers a clear blueprint: conversational AI that listens first, responds with intent, and builds trust over time.
Building these experiences requires more than models; it requires real-time infrastructure designed for expressive, low-latency interaction. Explore how Agora’s Conversational AI Engine delivers a complete, production-ready stack for voice, video, and AI-driven conversations through a demo or hands-on evaluation.
Convo AI World Japan is part of Agora’s Convo AI World Event Series, bringing together product leaders, developers, and AI innovators to connect on conversational AI implementation.
