ByteDance adds real-time video calls to Doubao

ByteDance, the parent company of TikTok, has introduced a real-time video call feature for its AI chatbot Doubao.

Users can now interact with the AI through video by activating their smartphone’s camera during a voice call, as announced on Doubao’s WeChat account on May 25, 2025.

This new feature allows Doubao to assist users in various situations.

These include providing museum tours, offering gardening advice, suggesting recipes while grocery shopping, and interpreting charts and videos.

🔗 Source: South China Morning Post

🧠 Food for thought

1️⃣ China’s generative AI landscape reflects years of strategic national investment

ByteDance’s Doubao video call upgrade exemplifies China’s rapid progress in multimodal AI, which stems from a long-term national strategy.

The Chinese government’s “New Generation AI Development Plan” launched in 2017 set an ambitious goal to create a $150 billion national AI industry by 2030, with a three-stage roadmap for global leadership1.

This coordinated approach involved designating tech giants as national champions in AI innovation, while providing substantial government funding that was projected to surpass private venture capital by 20162.

The current competition between ByteDance’s Doubao (107 million monthly active users) and Alibaba’s Quark (149 million monthly active users) represents the commercial culmination of this strategic investment3.

China’s advantage in AI development has been partly fueled by its vast consumer database, which provides extensive data for training sophisticated AI models capable of handling complex visual reasoning tasks like those in Doubao’s new video function2.

2️⃣ Multimodal capabilities represent the new competitive frontier in consumer AI

The real-time video call function in Doubao demonstrates how multimodal capabilities (combining visual, audio, and text processing) have become the key battleground for consumer AI apps.

ByteDance’s approach with Doubao parallels recent developments from competitors, including Alibaba’s introduction of its Qwen2.5-Omni-7B multimodal AI model in March and OpenAI’s GPT-4o update that significantly boosted ChatGPT’s user numbers with enhanced image generation capabilities3.

This pattern of multimodal feature competition shows how AI companies are racing to create more intuitive and natural human-computer interfaces that can understand and respond to the world as humans do.