Generative graphical explanations using large language models in AI-based services
Methods, systems, and computer-readable storage media for receiving a query from a digital assistant of an enterprise system, retrieving data that is responsive to the query from a data management system, inputting a first few-shot prompt to a LLM, and determining, in response to the first few-shot prompt, that a graphical representation of the data is to be generated, and in response: inputting a second few-shot prompt and a third few-shot prompt to the LLM, receiving code from the LLM responsive to the third few-shot prompt, and executing the code to render the graphical representation with the digital assistant.
1. A computer-implemented method for selectively generating graphical representations with digital assistants in enterprise systems, the method being executed by one or more processors and comprising:
receiving a query from a digital assistant of an enterprise system;
retrieving data that is responsive to the query from a data management service;
inputting a first few-shot prompt to a LLM, the first few-shot prompt being generated using a first prompt template; and
determining, in response to the first few-shot prompt, that a graphical representation of the data is to be generated, and in response:
inputting a second few-shot prompt and a third few-shot prompt to the LLM, the second few-shot prompt being generated using a second prompt template that is different from the first prompt template and the third few-shot prompt being generated using a third prompt template that is different from the second prompt template,
receiving code from the LLM responsive to the third few-shot prompt, and
executing the code to render the graphical representation with the digital assistant.
2. The method of claim 1 , wherein the graphical representation is rendered within a popover container.
3. The method of claim 2 , wherein the digital assistant communicates with the popover container using remote procedure calls (RPCs) to execute the code and render the graphical representation.
4. The method of claim 1 , further comprising displaying explanatory text within the digital assistant, the explanatory text being provided from the LLM in response to the second few-shot prompt and providing an explanation for the data and is responsive to the query.
5. The method of claim 1 , wherein determining, in response to the first few-shot prompt, that a graphical representation of the data is to be generated comprises generating a response by the LLM responsive to the first few-shot prompt, the response indicating that a graphical representation of the data is to be generated.
6. The method of claim 1 , wherein the digital assistant is provided in an application.
7. The method of claim 1 , wherein the data management service is agnostic to multiple applications executing digital assistants.
8. A non-transitory computer-readable storage medium coupled to one or more processors and having instructions stored thereon which, when executed by the one or more processors, cause the one or more processors to perform operations for selectively generating graphical representations with digital assistants in enterprise systems, the operations comprising:
receiving a query from a digital assistant of an enterprise system;
retrieving data that is responsive to the query from a data management service;
inputting a first few-shot prompt to a LLM, the first few-shot prompt being generated using a first prompt template; and
determining, in response to the first few-shot prompt, that a graphical representation of the data is to be generated, and in response:
inputting a second few-shot prompt and a third few-shot prompt to the LLM, the second few-shot prompt being generated using a second prompt template that is different from the first prompt template and the third few-shot prompt being generated using a third prompt template that is different from the second prompt template,
receiving code from the LLM responsive to the third few-shot prompt, and
executing the code to render the graphical representation with the digital assistant.
9. The non-transitory computer-readable storage medium of claim 8 , wherein the graphical representation is rendered within a popover container.
10. The non-transitory computer-readable storage medium of claim 9 , wherein the digital assistant communicates with the popover container using remote procedure calls (RPCs) to execute the code and render the graphical representation.
11. The non-transitory computer-readable storage medium of claim 8 , wherein operations further comprise displaying explanatory text within the digital assistant, the explanatory text being provided from the LLM in response to the second few-shot prompt and providing an explanation for the data and is responsive to the query.
12. The non-transitory computer-readable storage medium of claim 8 , wherein determining, in response to the first few-shot prompt, that a graphical representation of the data is to be generated comprises generating a response by the LLM responsive to the first few-shot prompt, the response indicating that a graphical representation of the data is to be generated.
13. The non-transitory computer-readable storage medium of claim 8 , wherein the digital assistant is provided in an application.
14. The non-transitory computer-readable storage medium of claim 8 , wherein the data management service is agnostic to multiple applications executing digital assistants.
15. A system, comprising:
a computing device; and
a computer-readable storage device coupled to the computing device and having instructions stored thereon which, when executed by the computing device, cause the computing device to perform operations for selectively generating graphical representations with digital assistants in enterprise systems, the operations comprising:
receiving a query from a digital assistant of an enterprise system;
retrieving data that is responsive to the query from a data management service;
inputting a first few-shot prompt to a LLM, the first few-shot prompt being generated using a first prompt template; and
determining, in response to the first few-shot prompt, that a graphical representation of the data is to be generated, and in response:
inputting a second few-shot prompt and a third few-shot prompt to the LLM, the second few-shot prompt being generated using a second prompt template that is different from the first prompt template and the third few-shot prompt being generated using a third prompt template that is different from the second prompt template,
receiving code from the LLM responsive to the third few-shot prompt, and
executing the code to render the graphical representation with the digital assistant.
16. The system of claim 15 , wherein the graphical representation is rendered within a popover container.
17. The system of claim 16 , wherein the digital assistant communicates with the popover container using remote procedure calls (RPCs) to execute the code and render the graphical representation.
18. The system of claim 15 , wherein operations further comprise displaying explanatory text within the digital assistant, the explanatory text being provided from the LLM in response to the second few-shot prompt and providing an explanation for the data and is responsive to the query.
19. The system of claim 15 , wherein determining, in response to the first few-shot prompt, that a graphical representation of the data is to be generated comprises generating a response by the LLM responsive to the first few-shot prompt, the response indicating that a graphical representation of the data is to be generated.
20. The system of claim 15 , wherein the digital assistant is provided in an application.