Making interactive video a medium anyone can reach for in the age of AI.
DOVER, Del., July 22, 2026 /PRNewswire/ — SigmaZ, a trailblazing AI lab aiming to redefine interface between humans and information, has introduced the product vision for Tap8, a platform designed to move AI interaction beyond today’s text and voice interfaces and one-way video playback, into an interactive information interface that users can click, question, modify and receive feedback from in real-time.
The company aims to revolutionize how technology supports everyday scenarios, transforming video content into an interactive experience. By empowering users to gain more detailed insight into the world around them, ask questions and take action directly through visuals, Tap8 seeks to overhaul the traditional methods of digital content interaction.
Founders Demo Tap8: Click. Ask. Watch It Respond. Live｜SigmaZ AI Lab
Tap8 Will Transform Flat, One-way Playback into a Living Interface
SigmaZ believes the next stage of AI interaction will move beyond text-based chat, allowing users to engage directly with visual content. Rather than relying on question-and-answer exchanges, Tap8 will enable users to explore, analyze and interact with information already embedded in images and videos, creating a more fluid experience for applications such as product discovery, fitness and education.
The platform is powered by a Diffusion Language Model (DLM) rather than a conventional autoregressive large language model. By refining outputs across an entire scene instead of generating content sequentially, the model is designed to deliver the speed, consistency and responsiveness needed for real-time interactive video experiences.
William Yang, co-founder and CEO of SigmaZ AI Lab, said the future of AI interaction will depend not only on model intelligence, but also on how effectively AI can communicate with humans. He believes visual interaction represents a revolutionary direction in making AI experiences more natural and intuitive.
“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.”
Derek Law, co-founder and CTO of SigmaZ AI Lab, added that the company aims to gradually expand Tap8’s capabilities and bring visual interaction to more everyday scenarios in the future, building upon the rapid ongoing development of AI-powered video.
“Tap8 is not simply about generating videos. Our vision is to explore how video can become an information interface that users can interact with. As the technology and product capabilities continue to evolve, we hope to help users engage with video more intuitively. 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.”
To enable interactive video experiences for everyday use, Tap8 is designed to address key challenges including response speed, interaction quality, generation stability and inference costs. SigmaZ is building Tap8 from the infrastructure layer, powered by an in-house-developed training framework built on the DLM approach.
DLM can support real-time-level latency while achieving leading visual code quality and inference costs approximately 1/20 of traditional DiT-based approaches.
Tap8 also incorporates SigmaZ AI Lab’s self-developed Vision-guided RSI (Recursive Self-Improvement) framework, which enables AI to evaluate and improve its own visual outputs. The framework was accepted this April at the ICLR 2026 RSI Workshop in Rio de Janeiro, providing strong third-party academic validation of the technology.
The system generates executable frontend code, renders the result into a visual output, uses a vision model to assess differences and feeds the feedback back into the model. This process enables continuous improvement in AI video generation through real visual feedback, supporting the future development of interactive video applications.
Full benchmark breakdown, methodology and appendix tables are up on our blog — read more here.
We Benchmarked TapVid Against Veo, Seedance & 5 Other AI Video Models — Here’s What Won
About SigmaZ AI Lab
SigmaZ is 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.
The company was co-founded by William Yang and Derek Law. Its core team brings together researchers, engineers and product leaders from AI teams including 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.
SigmaZ is building general-purpose interactive video infrastructure to transform video from one-way content into real-time generated visual interaction, with the goal of developing a next-generation information platform for everyday users. Users, developers, media, investors and partners interested in Tap8 can visit tap8.ai or sigmaz.io for more information.
Readers can also join the Tap8 waitlist to follow future product updates, demo releases and early access opportunities.
For media and partnership inquiries, please contact:
Frank Ge
frank.ge@sigmatrx.ai
