Computer session management using a language model

A computing device comprising:

• at least one processor;
• and
• a non-transitory computer-readable medium storing executable instructions that cause the at least one processor to:
• render, on the computing device, an interface to communicate with a language model;
• generate context data about computer activity of a user on the computing device, the context data being associated with a user session of the user and including information about a plurality of browser tabs opened on the computing device;
• condition the language model by transmitting a first prompt to the language model via a background request between the computing device and the language model, the first prompt including the context data associated with the user session;
• receive, via the interface, a user-generated query about a browser tab of the plurality of browser tabs;
• transmit a second prompt to the language model via an explicit request between the interface and the language model, the second prompt being associated with the user session and including the user-generated query;
• adjust a value of a context parameter of the language model based on whether a number of previous user-generated queries to the interface are received within a threshold period of time, the context parameter defining a threshold number of tokens used to generate a model response;
• and
• provide the model response on the interface of the language model, the model response being generated by the language model using the context data of the first prompt and the user-generated query of the second prompt.

A non-transitory computer-readable medium storing executable instructions that cause at least one processor to execute operations, the operations comprising:

• rendering, on a computing device, an interface to communicate with a language model;
• generating context data about computer activity of a user on the computing device, the context data being associated with a user session of the user and including information about a plurality of browser tabs opened on the computing device;
• conditioning the language model by transmitting a first prompt to the language model via a background request between the computing device and the language model, the first prompt including-based on the context data associated with the user session;
• receiving, via the interface of the language model, a user-generated query about a browser tab rendered on the computing device;
• transmitting a second prompt to the language model via an explicit request between the interface and the language model, the second prompt being associated with the user session and including the user-generated query;
• receiving, from the language model, a model response about the browser tab, the model response being generated by the language model using the context data of the first prompt and the user-generated query of the second prompt;
• and
• providing the model response on the interface of the language model.