Abstract
Conversational agents integrated into software applications can introduce user friction by providing text-based instructions that require users to perform manual navigation. This disclosure describes systems and methods where a language model can process a user's natural language input to generate a structured data payload, such as a Javascript object notation object, containing executable commands. A client-side application may intercept this payload and use a native execution engine to interpret the commands. This process can trigger direct user interface actions, for example, navigating a user to a specific screen or dynamically rendering an interactive native widget within the conversational flow. This approach may reduce navigational complexity and aid in feature discovery by translating a user's semantic intent into application functionality, offering an alternative to text-only instructional responses.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Aggar, Ankit and Mohan, Sujay, "Language Model Generation of Executable Commands for Native User Interface Control", Technical Disclosure Commons, ()
https://www.tdcommons.org/dpubs_series/12389