Method and system for supplementing messages with contextual content

1. A method comprising:

receiving, at a processor, a message from a sender to a user over a network, the message including a content item;

predicting, by the processor, a next action of the user based on the message by inputting the content item into a predictive model trained using historical user interface interactions performed by the user within a same user interface context as the predicted next action, the predictive model comprising a neural network and the next action comprising a data structure representing a user interaction performed with respect to a user interface (UI), the data structure including a type of action and one or more parameters defining the action;

selecting, by the processor, a template from a plurality of templates, each mapped to a respective action type based on the type of the next action;

generating, by the processor, an augmented content item using the selected template, the augmented content item including a control to perform the next action, the augmented content item generated by inserting the one or more parameters into respective placeholders within the selected template;

adding, by the processor, the augmented content item and the content item into respective sections of the UI in accordance with a predefined layout;

transmitting, by the processor, the UI to a client device of the user;

monitoring, by the processor, user interactions with the augmented content item in the UI; and

upon detecting a user interaction with the augmented content item: generating a new training vector comprising the content item of the received message and an action derived from the detected user interaction and updating the predictive model by retraining the neural network using the new training vector to refine prediction of next actions.

2. The method of claim 1 , wherein receiving a message from a sender to a user comprises receiving an email from the sender and wherein the content item includes one of text, image, or video content.

3. The method of claim 1 , further comprising training the neural network by generating a training dataset for the user, the training dataset generated using historical messages and historical actions performed within a forecast window relative to the historical messages.

4. The method of claim 1 , wherein the neural network outputs an action type and one or more parameters associated with the action type as the next action.

5. The method of claim 1 , wherein generating the augmented content item comprises identifying an external data source and generating an embedded module displaying data from the external data source.

6. The method of claim 5 , wherein the embedded module comprises an accelerated mobile pages (AMP) module.

7. The method of claim 1 , wherein adding the augmented content item and the content item into a UI comprises displaying the content item in a first pane of the UI and displaying the augmented content item alongside the first pane.

8. The method of claim 1 , wherein the augmented content item comprises one of a hyperlink, video, image, or text.

9. A non-transitory computer-readable storage medium for tangibly storing computer program instructions capable of being executed by a computer processor, the computer program instructions defining steps of:

receiving a message from a sender to a user over a network, the message including a content item;

predicting a next action of the user based on the message by inputting the content item into a predictive model trained using historical user interface interactions performed by the user within a same user interface context as the predicted next action, the predictive model comprising a neural network and the next action comprising a data structure representing a user interaction performed with respect to a user interface (UI), the data structure including a type of action and one or more parameters defining the action;

selecting, by the processor, a template from a plurality of templates, each mapped to a respective action type based on the type of the next action;

generating an augmented content item using the template, the augmented content item including a control to perform the next action, the augmented content item generated by inserting the one or more parameters into respective placeholders within the selected template;

adding the augmented content item and the content item into respective sections of the UI in accordance with a predefined layout;

transmitting the UI to a client device of the user;

monitoring user interactions with the augmented content item in the UI; and

upon detecting a user interaction with the augmented content item: generating a new training vector comprising the content item of the received message and an action derived from the detected user interaction and updating the predictive model by retraining the neural network using the new training vector to refine prediction of next actions.

10. The non-transitory computer-readable storage medium of claim 9 , the steps further comprising training the neural network by generating a training dataset for the user, the training dataset generated using historical messages and historical actions performed within a forecast window relative to the historical messages.

11. The non-transitory computer-readable storage medium of claim 9 , wherein the neural network outputs an action type and one or more parameters associated with the action type as the next action.

12. The non-transitory computer-readable storage medium of claim 9 , wherein generating the augmented content item comprises identifying an external data source and generating an embedded module displaying data from the external data source.

13. The non-transitory computer-readable storage medium of claim 12 , wherein the augmented content item comprises one of a hyperlink, video, image, or text.

14. A device comprising:

a processor; and

a storage medium for tangibly storing thereon logic for execution by the processor, the logic comprising instructions for:

receiving a message from a sender to a user over a network, the message including a content item;

predicting a next action of the user based on the message by inputting the content item into a predictive model trained using historical user interface interactions performed by the user within a same user interface context as the predicted next action, the predictive model comprising a neural network and the next action comprising a data structure representing a user interaction performed with respect to a user interface (UI), the data structure including a type of action and one or more parameters defining the action;

selecting, by the processor, a template from a plurality of templates, each mapped to a respective action type based on the type of the next action;

generating an augmented content item using the selected template, the augmented content item including a control to perform the next action, the augmented content item generated by inserting the one or more parameters into respective placeholders within the selected template;

adding the augmented content item and the content item into respective sections of the UI in accordance with a predefined layout;

transmitting the UI to a client device of the user;

monitoring user interactions with the augmented content item in the UI; and

upon detecting a user interaction with the augmented content item: generating a new training vector comprising the content item of the received message and an action derived from the detected user interaction and updating the predictive model by retraining the neural network using the new training vector to refine prediction of next actions.

15. The device of claim 14 , the instructions further comprising training the neural network by generating a training dataset for the user, the training dataset generated using historical messages and historical actions performed within a forecast window relative to the historical messages.

16. The device of claim 14 , wherein the neural network outputs an action type and one or more parameters associated with the action type as the next action.

17. The device of claim 14 , wherein generating the augmented content item comprises identifying an external data source and generating an embedded module displaying data from the external data source.