System and method for link-initiated user engagement and retention utilizing generative artificial intelligence
A system and methods for dynamic-link initiated user engagement and retention utilizing generative artificial intelligence. The system integrates a generalized generative AI, personalized to each user's context, generating pre-filled message to be displayed within a messaging application on a mobile device. It utilizes user profiles, interaction history, and deep link context to dynamically generate contextually relevant pre-filled messages. The system may employ an LSTM-based model with attention mechanisms for both timing and content prediction. It interfaces with the messaging app through an API, extracting deep link context and triggering AI-generated suggestions. This enables seamless, personalized follow-up messages accompanying deep links, fostering customer engagement and retention by providing timely and valuable interactions. Continuous monitoring and model updates ensure optimal performance and alignment with user preferences, ultimately enhancing user experience and long-term app engagement.
1 . A system for link-initiated user engagement and retention using generative artificial intelligence, comprising:
a first trained machine learning algorithm configured to generate an initial message comprising an attention mechanism configured to compute weights for portions of input data to capture contextual relevance, the initial message comprising a first deep link and a payload, the first deep link comprising a uniform resource identifier configured to direct a user to a specific location or content within a mobile application; and
a computing device comprising a processor, a memory, and a first plurality of programming instructions stored in the memory and operable on the processor, wherein the first plurality of programming instructions, when operating on the processor, cause the computing device to:
upon receipt of a user interaction from a mobile device substantially corresponding to a call-to-action and mobile device metadata, check if there is an active mobile device record associated with the mobile device;
when there is no active mobile device record associated with the mobile device, create and store an active mobile device record to be associated with the mobile device, the created active mobile device record comprising at least the mobile device metadata;
send a confirmation message to the mobile device responsive to the creation of the active mobile device record associated with the mobile device;
use the mobile device metadata and the active mobile device record as inputs into the first trained machine learning algorithm to determine the initial message when there is an active mobile device record associated with the mobile device; and
send the initial message to a messaging application on the mobile device via an application programming interface, wherein the payload is pre-populated into a message field of the messaging application.
2 . The system of claim 1 , further comprising a second trained machine learning algorithm configured to determine an optimal time to send a follow-up message and generate a content of the follow-up message; and
wherein the computing device is further configured to:
use the mobile device metadata and the active mobile device record as inputs into the second trained machine learning algorithm to determine the optimal time and generate the follow-up message; and
send the follow-up message to the messaging application on the mobile device at the determined optimal time.
3 . The system of claim 1 , wherein the payload comprises pre-populated message suitable for display in the messaging application.
4 . The system of claim 1 , wherein the active mobile device record comprises user profile information, an interaction history, and at least one user preference.
5 . The system of claim 1 , wherein the first machine learning algorithm is a recurrent neural network (RNN).
6 . The system of claim 5 , wherein the recurrent neural network is a long short-term memory (LSTM) RNN.
7 . The system of claim 5 , wherein the LSTM RNN is a generative artificial intelligence model.
8 . The system of claim 2 , wherein the second machine learning algorithm is an integrated model, wherein the integrated model comprises a time-series forecasting model and a natural language processing model.
9 . The system of claim 8 , wherein the integrated model is a generative artificial intelligence model.
10 . The system of claim 1 , wherein the dataset used to train first and second trained machine learning models comprises a plurality of active mobile device record information, interaction history, deep link context data, and device attributes.
11 . A method for link-initiated user engagement and retention using generative artificial intelligence, comprising the steps of:
training a first machine learning algorithm configured to generate an initial message comprising an attention mechanism configured to compute weights for portions of input data to capture contextual relevance, the initial message comprising a first deep link and a payload, the first deep link comprising a uniform resource identifier configured to direct a user to a specific location or content within a mobile application;
upon receipt of a user interaction from a mobile device substantially corresponding to a call-to-action and mobile device metadata, checking if there is an active mobile device record associated with the mobile device;
when there is no active mobile device record associated with the mobile device, creating and storing an active mobile device record to be associated with the mobile device, the created active mobile device record comprising at least the mobile device metadata;
sending a confirmation message to the mobile device responsive to the creation of the active mobile device record associated with the mobile device;
using the mobile device metadata and the active mobile device record as inputs into the first trained machine learning algorithm to determine the initial message when there is an active mobile device record associate with the mobile device; and
sending the initial message to a messaging application on the mobile device via an application programming interface, wherein the payload is pre-populated into a message field of the messaging application.
12 . The method of claim 11 , further comprising the steps of:
training a second machine learning algorithm configured to determine an optimal time to send a follow-up message and generate a content of the follow-up message;
using the mobile device metadata and the active mobile device record as inputs into the second trained machine learning algorithm to determine the optimal time and generate the follow-up message; and
sending the follow-up message to the messaging application on the mobile device at the determined optimal time.
13 . The method of claim 11 , wherein the payload comprises pre-populated message suitable for display in the messaging application.
14 . The method of claim 11 , wherein the active mobile device record further comprises user profile information, an interaction history, and at least one user preference.
15 . The method of claim 11 , wherein the first machine learning algorithm is a recurrent neural network (RNN).
16 . The method of claim 15 , wherein the recurrent neural network is a long short-term memory (LSTM) RNN.
17 . The method of claim 15 , wherein the LSTM RNN is a generative artificial intelligence model.
18 . The method of claim 12 , wherein the second machine learning algorithm is an integrated model, wherein the integrated model comprises a time-series forecasting model and a natural language processing model.
19 . The method of claim 18 , wherein the integrated model is a generative artificial intelligence model.
20 . The method of claim 11 , wherein the dataset used to train first and second trained machine learning models comprises a plurality of active mobile device record information, interaction history, deep link context data, and device attributes.