MUMBAI, India, Feb. 27 -- Intellectual Property India has published a patent application (202646017881 A) filed by Voicemonk Inc., Sunnyvale, U.S.A., on Feb. 17, for 'method and system for autonomous user interface orchestration using world action models and structured protocols.'
Inventor(s) include Jagadeshwar Nomula; and Vinesh Gudla.
The application for the patent was published on Feb. 27, under issue no. 09/2026.
According to the abstract released by the Intellectual Property India: "A virtual agent disclosed in parent patent application no. 201644039413 is improved with a method and system enabling an artificial intelligence agent to interact with a user interface. A Virtual Agent Server receives a natural language instruction and stores a Structured Representation encoded according to a standardized tool. A ReAct Planner generates a Proposed Action or Proposed Action Sequence. Prior to OS Event Injection, a World Action Model predicts a next interface state in latent space, and a verifier permits or blocks execution based on satisfaction of an expected postcondition and absence of an error condition. Screen perception may use a JEPA Encoder to generate Masked Latents and Action Prototype Heads to localize UI elements and classify Interaction Types. An execution policy may be optimized using Group Relative Policy Optimization using rewards derived from Task Success and/or an Interaction Friction Metric without a critic network."
Disclaimer: Curated by HT Syndication.
