Systems and methods for automatic generation of datasets for record objects using machine learning architectures

1 . A system comprising:

memory; and

one or more processors configured by machine-readable instructions stored in the memory to:

identify a plurality of electronic activities matched with a record object of a customer relationship management (CRM) system;

generate, by inputting data corresponding to the plurality of electronic activities into one or more large language models, a first set of text strings;

selecting a subset of the first set of text strings;

generating, by inputting the subset of the first set of text strings in the one or more large language models, a subset summary for the subset of the first set of text strings into the first set of text strings;

store, in one or more data structures, a first association between the first set of text strings including the subset summary for the subset of the first set of text strings and the record object based on the match between the record object and the plurality of electronic activities;

identify a second electronic activity that is matched with the record object subsequent to generating the first set of text strings;

generate, using the one or more large language models, responsive to determining that the second electronic activity is matched with the record object or a time condition is satisfied, a second set of text strings based on data corresponding to the second electronic activity and the first set of text strings; and

store, in the one or more data structures, a second association between the second set of text strings and the record object.

2 . The system of claim 1 , wherein the one or more processors are configured by the machine-readable instructions to generate the first set of text strings by:

generating, for each of the plurality of electronic activities, an activity summary of the electronic activity by inputting data of the electronic activity into the one or more large language models.

3 . The system of claim 1 , wherein the one or more processors are configured by the machine-readable instructions to select the subset of the first set of text strings by:

selecting the subset of the first set of text strings based on each text string of the subset corresponding to a common time interval; or

selecting the subset of the first set of text strings responsive to determining a size of the subset of text strings has reached at least a defined value.

4 . The system of claim 1 , wherein the one or more processors are configured by the machine-readable instructions to select the subset of the first set of text strings by:

selecting the subset of the first set of text strings responsive to determining each text string of the subset corresponds to a common context or identifies a common individual.

5 . The system of claim 1 , wherein the one or more processors are configured by the machine-readable instructions to generate the subset summary for the subset of text strings by:

assigning a summary identifier to the subset summary for the subset of text strings and a text string identifier to each of the text strings of the subset; and

including the summary identifier and the text string identifier for each text string of the subset in the subset summary for the subset of the first set of text strings.

6 . The system of claim 1 , wherein the one or more processors are further configured by the machine-readable instructions to:

discard the plurality of electronic activities subsequent to generating the first set of text strings.

7 . The system of claim 1 , wherein the one or more processors are further configured by the machine-readable instructions to:

receive, via a chat interface, a request including a natural language query requesting information regarding the record object, the request comprising an identification of the record object; and

generate, using the one or more large language models, a response to the natural language query containing the requested information using the identification of the record object and the stored association between the record object and the first set of text strings or the second set of text strings generated prior to the query.

8 . The system of claim 7 , wherein the one or more processors are further configured by the machine-readable instructions to generate the response to the natural language query by:

retrieving one or more text strings from the second set of text strings and one or more object field-value pairs from the record object; and

generating the response by inputting the retrieved one or more text strings and one or more object field-value pairs of the record object into the one or more large language models.

9 . The system of claim 1 , wherein the one or more processors are further configured by the machine-readable instructions to generate the second set of text strings by:

generating, by inputting data of the second electronic activity into the one or more large language models, a second text string summarizing the data of the second electronic activity; and

inserting the second text string into the first set of text strings.

10 . The system of claim 1 , wherein the one or more processors are configured by the machine-readable instructions to:

store the record object of the CRM system as a shadow record object of a shadow CRM system stored in memory, the shadow CRM system mapped to an external CRM system stored in a remote computing system, and

wherein the one or more processors are configured to identify the plurality of electronic activities matched with the record object of the CRM system by identifying the plurality of electronic activities matched with the shadow record object of the shadow CRM system.

11 . A computer-implemented method, comprising:

identifying, by one or more processors, a plurality of electronic activities matched with a record object of a customer relationship management (CRM) system;

generating, by the one or more processors, by inputting data corresponding to the plurality of electronic activities into one or more large language models, a first set of text strings;

selecting, by the one or more processors, a subset of the first set of text strings;

generating, by the one or more processors, by inputting the subset of the first set of text strings in the one or more large language models, a subset summary for the subset of the first set of text strings into the first set of text strings;

storing, by the one or more processors in one or more data structures, a first association between the first set of text strings including the subset summary for the subset of the first set of text strings and the record object based on the match between the record object and the plurality of electronic activities;

identifying, by the one or more processors, a second electronic activity that is matched with the record object subsequent to generating the first set of text strings;

generating, by the one or more processors, using the one or more large language models, responsive to determining that the second electronic activity is matched with the record object or a time condition is satisfied, a second set of text strings based on data corresponding to the second electronic activity and the first set of text strings; and

storing, by the one or more processors, in the one or more data structures, a second association between the second set of text strings and the record object.

12 . The method of claim 11 , wherein generating the first set of text strings comprises:

generating, by the one or more processors for each of the plurality of electronic activities, an activity summary of the electronic activity by inputting data of the electronic activity into the one or more large language models.

13 . The method of claim 11 , wherein selecting the subset of the first set of text strings comprises:

selecting, by the one or more processors, the subset of the first set of text strings based on each text string of the subset corresponding to a common time interval; or

selecting, by the one or more processors, the subset of the first set of text strings responsive to determining a size of the subset of text strings has reached at least a defined value.

14 . The method of claim 11 , wherein selecting the subset of the first set of text strings comprises:

selecting, by the one or more processors, the subset of the first set of text strings responsive to determining each text string of the subset corresponds to a common context or identifies a common individual.

15 . The method of claim 11 , wherein generating the subset summary for the subset of text strings comprises:

assigning, by the one or more processors, a summary identifier to a subset summary for the subset of text strings and a text string identifier to each of the text strings of the subset; and

including, by the one or more processors, the summary identifier and the text string identifier for each text string of the subset in the subset summary for the subset of the first set of text strings.

16 . The method of claim 11 , further comprising:

discarding, by the one or more processors, the plurality of electronic activities subsequent to generating the first set of text strings.