Generating smart topics for video calls using a large language model and a context transformer engine
The present disclosure relates to systems, non-transitory computer-readable media, and methods for utilizing a context transformer engine, a smart topic agent, and a large language model to generate a smart topic output. In particular, in one or more embodiments, the disclosed systems generate a smart topic output from a transcript of a video call. In some embodiments, the disclosed systems provide a smart topic interface that provides the smart topic output on a client device and receives selections of smart topic elements. In one or more embodiments, the disclosed systems generate a combined smart topic from transcripts of video calls in which client devices that participated are associated with a collaborating user account group.
1 . A computer-implemented method comprising:
determining a collaborating user account group comprising user accounts of a content management system collaborating together on one or more content items corresponding to a smart topic;
obtaining a first transcript from a first video call that includes first video call data associated with a smart topic captured from a first client device associated with the collaborating user account group;
obtaining a second transcript from a second video call that includes second video call data associated with the smart topic captured from a second client device associated with the collaborating user account group;
receiving, from the first client device, a selection of a smart topic element corresponding to the smart topic; and
in response to the selection of the smart topic element, generating a combined smart topic output from the first transcript and the second transcript by utilizing a smart topic agent, a context transformer engine, and a large language model.
2 . The computer-implemented method of claim 1 , wherein generating the combined smart topic output comprises:
identifying, utilizing the context transformer engine and the smart topic agent, a first portion of the first transcript comprising subject matter associated with the smart topic and a second portion of the second transcript comprising subject matter associated with the smart topic; and
providing the first portion of the first transcript and the second portion of the second transcript to the large language model to generate the combined smart topic output.
3 . The computer-implemented method of claim 1 , further comprising:
receiving, from the first client device, a text input indicating subject matter for the combined smart topic output; and
generating the combined smart topic output based on the text input.
4 . The computer-implemented method of claim 1 , wherein obtaining the first transcript from the first video call comprises:
receiving, from the first client device, an indication of a user interaction selecting a capture element within a smart topic interface;
based on receiving the indication of the user interaction, capturing the first video call data from the first client device during the first video call; and
generating the first transcript from the first video call data from the first client device.
5 . The computer-implemented method of claim 1 , further comprising:
obtaining the first transcript from the first video call based on receiving, from the first client device, a selection of a first selectable option to share transcript data; and
obtaining the second transcript from the second video call based on receiving, from the second client device, a selection of a second selectable option to share transcript data.
6 . The computer-implemented method of claim 1 , further comprising:
identifying that a first portion of the first transcript and a second portion of the second transcript comprise subject matter corresponding to the smart topic; and
displaying the first portion of the first transcript and the second portion of the second transcript within a smart topic interface presented on the first client device.
7 . The computer-implemented method of claim 1 , further comprising:
generating a suggested combined smart topic based on one or more of: identifying that the first transcript and the second transcript comprise related subject matter, receiving indications of user interactions with the first transcript or the second transcript, or previous text input from the first client device to generate a previous smart topic;
providing, for display within a smart topic interface presented on the first client device, a suggested smart topic element corresponding to the suggested combined smart topic; and
based on receiving a selection of a suggested combined smart topic element corresponding to the suggested combined smart topic, generating the suggested combined smart topic.
8 . The computer-implemented method of claim 1 , further comprising:
identifying that one or more additional content items stored for a user account associated with the collaborating user account group are associated with the smart topic; and
providing, within a smart topic interface on the first client device, the combined smart topic output and an option to view the one or more additional content items.
9 . The computer-implemented method of claim 1 , further comprising:
identifying that a third client device that participated in the first video call is not associated with the collaborating user account group;
receiving, from the third client device, a selection of an option to utilize transcript data from the third client device when generating the combined smart topic output; and
generating the combined smart topic output based on receiving the selection of the option to utilize transcript data from the third client device.
10 . A non-transitory computer-readable medium storing instructions that, when executed by at least one processor, cause a computer system to:
determine a collaborating user account group comprising user accounts of a content management system collaborating together on one or more content items corresponding to a smart topic;
obtain a first transcript from a first video call that includes first video call data associated with a smart topic captured from a first client device associated with the collaborating user account group;
obtain a second transcript from a second video call that includes second video call data associated with the smart topic captured from a second client device associated with the collaborating user account group;
obtain application data from an application associated with a user account of the collaborating user account group and separate from the first video call data;
receive, from the first client device, a selection of a smart topic element corresponding to the smart topic; and
in response to the selection of the smart topic element, generate a combined smart topic output from the first transcript, the second transcript, and the application data by utilizing a smart topic agent, a context transformer engine, and a large language model.
11 . The non-transitory computer-readable medium of claim 10 , further comprising instructions that, when executed by the at least one processor, cause the computer system to:
identify, utilizing the context transformer engine and the smart topic agent, a first portion of the first transcript comprising subject matter associated with the smart topic and a second portion of the second transcript comprising subject matter associated with the smart topic; and
provide the first portion of the first transcript and the second portion of the second transcript to the large language model to generate the combined smart topic output.
12 . The non-transitory computer-readable medium of claim 10 , further comprising instructions that, when executed by the at least one processor, cause the computer system to generate, utilizing the context transformer engine, the smart topic agent as executable computer code for performing a process for the smart topic, wherein the smart topic agent is specific to a first particular portion of the first transcript and a second particular portion of the second transcript.
13 . The non-transitory computer-readable medium of claim 10 , further comprising instructions that, when executed by the at least one processor, cause the computer system to:
compare the first transcript, the second transcript, and the application data;
based on comparing the first transcript, the second transcript, and the application data, determine that the first transcript, the second transcript, and the application data comprise related subject matter; and
generate the combined smart topic output based on determining that the first transcript, the second transcript, and the application data comprise related subject matter.
14 . The non-transitory computer-readable medium of claim 10 , further comprising instructions that, when executed by the at least one processor, cause the computer system to:
receive, from the first client device, a text input indicating subject matter for the combined smart topic output;
generate the combined smart topic output based on the text input; and
provide the combined smart topic output for display within a smart topic interface presented on the first client device.
15 . The non-transitory computer-readable medium of claim 10 , further comprising instructions that, when executed by the at least one processor, cause the computer system to:
generate a suggested combined smart topic based on one or more of: identifying that the first transcript and the second transcript comprise related subject matter, receiving user interactions with the first transcript or the second transcript, or previous text input from the first client device to generate a previous smart topic;
provide, for display within a smart topic interface presented on the first client device, a suggested combined smart topic element corresponding to the suggested combined smart topic; and
based on receiving a selection of the suggested combined smart topic element, generate the suggested combined smart topic.
16 . A system comprising:
at least one processor; and
at least one non-transitory computer-readable storage medium storing instructions that, when executed by the at least one processor, cause the system to:
determine a collaborating user account group comprising user accounts of a content management system collaborating together on one or more content items corresponding to a smart topic;
obtain a first transcript from a first video call that includes first video call data associated with a smart topic captured from a first client device associated with the collaborating user account group;
obtain a second transcript from a second video call that includes second video call data associated with the smart topic captured from a second client device associated with the collaborating user account group;
receive, from the first client device, a selection of a smart topic element corresponding to the smart topic; and
in response to the selection of the smart topic element, generate a combined smart topic output from the first transcript and the second transcript by utilizing a smart topic agent, a context transformer engine, and a large language model.
17 . The system of claim 16 , further comprising instructions that, when executed by the at least one processor, cause the system to:
identify, utilizing the context transformer engine and the smart topic agent, a first portion of the first transcript comprising subject matter associated with the smart topic and a second portion of the second transcript comprising subject matter associated with the smart topic; and
provide the first portion of the first transcript and the second portion of the second transcript to the large language model to generate the combined smart topic output.
18 . The system of claim 16 , further comprising instructions that, when executed by the at least one processor, cause the system to:
receive, from the first client device, a text input indicating subject matter for the combined smart topic output;
generate the combined smart topic output based on the text input; and
provide the combined smart topic output for display within a smart topic interface presented on the first client device.
19 . The system of claim 16 , further comprising instructions that, when executed by the at least one processor, cause the system to:
generate a suggested combined smart topic based on one or more of: identifying that the first transcript and the second transcript comprise related subject matter, receiving an indication of a user interaction with the first transcript or the second transcript, or previous text input from the first client device to generate a previous smart topic;
provide, for display within a smart topic interface presented on the first client device, a suggested smart topic element corresponding to the suggested combined smart topic; and
based on receiving a selection of a suggested combined smart topic element corresponding to the suggested combined smart topic, generating the suggested combined smart topic.
20 . The system of claim 16 , further comprising instructions that, when executed by the at least one processor, cause the system to:
receive, from the first client device, a text input of a search query associated with the smart topic in a transcript searching interface; and
based on receiving the text input of the search query, provide, to the first client device and in the transcript searching interface, a first portion of the first transcript and a second portion of the second transcript that correspond to the smart topic.