Runway's Solaris generates an interactive interface as a stream of images. Starting from a scene, it uses clicks, drags and other inputs to shape subsequent frames. A separate LLMA large language model — the neural network behind tools like Claude and ChatGPT, trained on huge amounts of text to predict what comes next.Full definition → interprets requests and supplies instructions that guide the visual model. This lets designers describe how a scene should respond without programming each interaction individually. Runway's technical account explains that division between deciding what happens and rendering it.
The approach remains experimental: Runway identifies readable text, consistency during long sessions and compatibility with assistive technologies as unresolved problems. Those constraints matter for interfaces people must reliably read and operate.
Terms in this piece · Glossary
LLM — A large language model — the neural network behind tools like Claude and ChatGPT, trained on huge amounts of text to predict what comes next.
