SigmaZ AI Lab has introduced Tap8, a platform designed to transform video from passive content into interactive visual interfaces where users can click, ask questions, and receive real-time responses. Rather than watching videos play out linearly, users will be able to explore, analyze, and interact with information embedded directly in visual content, marking a significant shift in how people engage with AI-generated media.

What Makes Tap8 Different From Today's AI Video Tools?

The core innovation behind Tap8 lies in its underlying technology architecture. Instead of using conventional autoregressive large language models (LLMs), which generate content one piece at a time, Tap8 is powered by a Diffusion Language Model (DLM). This approach refines outputs across an entire scene simultaneously, enabling the speed and responsiveness required for real-time interaction.

The performance gains are substantial. SigmaZ claims that DLM-based inference costs approximately 1/20th of traditional approaches used by competing systems, while maintaining leading visual quality. This efficiency matters because it determines whether interactive video can scale beyond research demos into everyday applications.

"AI's bottleneck is not intelligence, it is the bandwidth between AI and humans. Text and voice remain important, but they are not always the most natural entry points for interaction. We believe future AI experiences should allow users to ask questions, explore and interact directly with visual content," said William Yang, co-founder and CEO of SigmaZ AI Lab.

William Yang, Co-founder and CEO, SigmaZ AI Lab

How Does Tap8 Improve Its Own Video Quality?

• Vision-Guided Feedback Loop: Tap8 incorporates SigmaZ's self-developed Vision-guided RSI (Recursive Self-Improvement) framework, which enables the AI to evaluate and improve its own visual outputs automatically without human intervention.
• Real-Time Assessment: The system generates executable frontend code, renders the result into a visual output, and uses a vision model to assess differences between the intended and actual result.
• Continuous Learning: Feedback from the vision assessment is fed back into the model, supporting continuous improvement in video generation quality through real visual feedback rather than relying solely on human annotation.

This Vision-guided RSI framework received academic validation when it was accepted at the ICLR 2026 RSI Workshop in Rio de Janeiro this April, providing third-party credibility to SigmaZ's technical approach.

What Real-World Applications Could Tap8 Enable?

SigmaZ envisions Tap8 transforming several everyday scenarios. Product discovery could shift from static images and descriptions to interactive videos where users click on items to learn more. Fitness applications could allow users to ask questions about form or technique while watching instructional videos. Education could become more dynamic, with students able to explore concepts visually and ask clarifying questions in real-time.

The company's vision extends beyond video generation itself. Co-founder Derek Law emphasized that Tap8 represents a fundamental shift in how video functions as a medium. "Tap8 is not simply about generating videos. Our vision is to explore how video can become an information interface that users can interact with," Law stated. He compared this moment to past media revolutions, where the underlying substrate changed before new mediums emerged.

Derek Law

"Every past media revolution rewrote the substrate first, and a new medium grew out of it. We think AI is the next substrate like that," explained Derek Law, co-founder and CTO of SigmaZ AI Lab.

Derek Law, Co-founder and CTO, SigmaZ AI Lab

Who Is Behind SigmaZ and What's Their Background?

SigmaZ was co-founded by William Yang and Derek Law, bringing together a core team of researchers, engineers, and product leaders from major AI organizations. The team includes veterans from Amazon AGI, Alibaba AI Lab, ByteDance AI, and Baidu. The company has received investment from Lanchi Ventures, Yunqi Partners, Decent Capital, and Xiaoxiao Fund, among others.

The company positions itself as an interactive intelligence product lab focused on building visual interaction experiences for the AGI era. Unlike traditional AI interaction through text and voice, SigmaZ believes visual interaction will become an important medium for future human-AI interaction and information delivery.

Tap8 is currently accepting signups for its waitlist, with the company planning to release demos and offer early access opportunities to interested users, developers, media partners, and investors. The platform represents one of the first serious attempts to move AI video generation beyond one-way content creation into genuinely interactive visual experiences.