USING LARGE LANGUAGE MODELS TO GENERATE ELECTRONIC MESSAGES ASSOCIATED WITH CONTENT ITEMS
A system automatically generates electronic messages based on a plurality of signals received in a content management platform. The system receives, at a user interface generated by a computer system for display by a user device, an input to generate an electronic message. The system processes a plurality of signals retrieved based on the received input and generates a prompt to a large language model (LLM) to cause the LLM to generate content for the electronic message, where the prompt is generated at least in part based on the processing of the plurality of signals. The system populates, into the user interface, content for the electronic message that is generated based on output by the LLM.
1 . A computer-implemented method comprising:
receiving, at a user interface generated by a computer system for display by a user device, an input to generate an electronic message, wherein the input to generate the electronic message includes an identification of one or more digital content items to be associated with the electronic message;
processing, by the computer system, a plurality of signals retrieved based on the received input,
wherein the plurality of signals include content metadata associated with the one or more digital content items;
generating, by the computer system, a prompt to a large language model (LLM) to cause the LLM to generate content for the electronic message, wherein the prompt is generated at least in part based on the processing of the plurality of signals and instructs the LLM to generate a summary of the one or more digital content items for inclusion in the electronic message based on the content metadata;
populating, into the user interface, content for the electronic message that is generated based on output by the LLM; and
generating a transmissible payload that includes the content for the electronic message and the one or more digital content items attached to or linked within the transmissible payload.
2 . The computer-implemented method of claim 1 , wherein the plurality of signals include an identifier of a sender of the electronic message, and wherein the identifier of the sender is retrieved based on a login to a user account prior to the input to generate the electronic message being received.
3 . The computer-implemented method of claim 1 , wherein the plurality of signals include an identifier of a recipient for the electronic message, and wherein the identifier of the recipient is input at the user interface in association with the input to generate the electronic message.
4 . The computer-implemented method of claim 3 , wherein processing the plurality of signals comprises processing the identifier of the recipient to select message features for the electronic message, and wherein generating the prompt to the LLM comprises instructing the LLM to generate the content for the electronic message using the selected message features.
5 . The computer-implemented method of claim 1 , wherein the plurality of signals further include use data that characterizes user activity associated with the one or more digital content items.
6 . The computer-implemented method of claim 1 , wherein the plurality of signals include performance metrics associated with a plurality of prior electronic messages transmitted to respective recipients, and wherein processing the performance metrics comprises:
using the performance metrics to modify a prompt generation model; and
using the modified prompt generation model to generate the prompt to the LLM.
7 . The computer-implemented method of claim 6 , wherein the performance metrics characterize one or more of:
a number of the prior electronic messages that were read by the respective recipients;
a number of content items attached to the prior messages that were accessed by the respective recipients after the prior messages were transmitted to the respective recipients; or
a number of actions taken by the respective recipients after the prior messages were transmitted to the respective recipients.
8 . The computer-implemented method of claim 1 , wherein the plurality of signals include an identification of a category of electronic message to be sent, and wherein the prompt to the LLM specifies the identified category.
9 . The computer-implemented method of claim 1 , wherein populating the content for the electronic message into the user interface comprises generating a subject line and message body for the electronic message, and wherein the transmissible payload includes the subject line, message body.
10 . A non-transitory computer-readable storage medium storing executable computer program instructions, the computer program instructions when executed by one or more processors of a system causing the system to:
receive, at a user interface generated by the system for display by a user device, an input to generate an electronic message;
process a plurality of signals retrieved based on the received input;
generate a prompt to a large language model (LLM) to cause the LLM to generate content for the electronic message, wherein the prompt is generated at least in part based on the processing of the plurality of signals; and
populate, into the user interface, content for the electronic message that is generated based on output by the LLM.
11 . The non-transitory computer-readable storage medium of claim 10 , wherein the plurality of signals include an identifier of a sender of the electronic message, and wherein the identifier of the sender is retrieved based on a login to a user account prior to the input to generate the electronic message being received.
12 . The non-transitory computer-readable storage medium of claim 10 , wherein the plurality of signals include an identifier of a recipient for the electronic message, and wherein the identifier of the recipient is input at the user interface in association with the input to generate the electronic message.
13 . The non-transitory computer-readable storage medium of claim 10 :
wherein the input to generate the electronic message includes an identification of one or more digital content items to be associated with the electronic message;
wherein the plurality of signals include content metadata associated with the one or more digital content items; and
wherein the prompt instructs the LLM to generate a summary of the one or more digital content items for inclusion in the electronic message based on the content metadata.
14 . The non-transitory computer-readable storage medium of claim 10 , wherein the plurality of signals include performance metrics associated with a plurality of prior electronic messages transmitted to respective recipients, and wherein processing the performance metrics comprises:
using the performance metrics to modify a prompt generation model; and
using the modified prompt generation model to generate the prompt to the LLM.
15 . The non-transitory computer-readable storage medium of claim 10 , wherein the instructions when executed further cause the system to:
generating, by the computer system, a transmissible payload for the electronic message, wherein the transmissible payload includes the content for the electronic message and one or more content items attached to or linked within the transmissible payload.
16 . A data processing system, comprising:
one or more processors; and
one or more non-transitory computer-readable storage media storing executable computer program instructions, the computer program instructions when executed by the one or more processors cause the data processing system to:
receive, at a user interface generated by the data processing system for display by a user device, an input to generate an electronic message;
process a plurality of signals retrieved based on the received input;
generate a prompt to a large language model (LLM) to cause the LLM to generate content for the electronic message, wherein the prompt is generated at least in part based on the processing of the plurality of signals; and
populate, into the user interface, content for the electronic message that is generated based on output by the LLM.
17 . The data processing system of claim 16 , wherein the plurality of signals include an identifier of a recipient for the electronic message, and wherein the identifier of the recipient is input at the user interface in association with the input to generate the electronic message.
18 . The data processing system of claim 16 :
wherein the input to generate the electronic message includes an identification of one or more digital content items to be associated with the electronic message;
wherein the plurality of signals include content metadata associated with the one or more digital content items; and
wherein the prompt instructs the LLM to generate a summary of the one or more digital content items for inclusion in the electronic message based on the content metadata.
19 . The data processing system of claim 16 , wherein the plurality of signals include performance metrics associated with a plurality of prior electronic messages transmitted to respective recipients, and wherein processing the performance metrics comprises:
using the performance metrics to modify a prompt generation model; and
using the modified prompt generation model to generate the prompt to the LLM.
20 . The data processing system of claim 16 , wherein the instructions when executed further cause the system to:
generating, by the computer system, a transmissible payload for the electronic message, wherein the transmissible payload includes the content for the electronic message and one or more content items attached to or linked within the transmissible payload.