Back to updates
Lab NoteFeatured
MoLing: why visible state came first
Before connecting a production model, we built listening, thinking, retrieving and answering into a visible visual language.
Most conversational interfaces have two states: waiting for input, and a spinner. A user cannot tell whether the system is retrieving or generating, or whether an answer has any grounding.
MoLing splits interaction into four states and gives each its own visual expression. The immediate benefit is that when an answer disappoints, the user can tell roughly which stage went wrong.
This is an interface-layer experiment and does not treat one integration result as a model benchmark. MoLing on this site runs in integration mode, with text and live voice sharing one Mo-Voice session.