Automated translations for autonomous chat agents

1 . A system to reduce processing cycles associated with execution of models used by autonomous chat agents, comprising:

one or more processors and a memory; and

machine-readable instructions stored in the memory that, when executed by the one or more processors, cause the one or more processors to at least:

receive, from a service executed by the one or more processors, a request to update an autonomous chat agent, wherein the autonomous chat agent uses a workflow to process queries in a first language, wherein the workflow includes: (i) at least one intent, (ii) one or more responses, (iii) training data, and (iv) a machine learning model configured to recognize the at least one intent in the first language and provide one or more responses based on the at least one intent;

generate, responsive to receipt of the request, a translated workflow from the workflow, wherein the translated workflow corresponds to a second language and to generate the translated workflow, the one or more processors:

send, to a machine translation service, a translation request that specifies the second language, the translation request comprising the training data having words or phrases of the workflow in the first language, wherein the words or phrases correspond to the at least one intent of the workflow;

receive, from the machine translation service responsive to the translation request, translated training data in the second language that is generated, by the machine translation service, from a translation of the training data in the first language, wherein the translated training data indicates words or phrases in the second language that relate to the at least one intent of the workflow;

and

train a translated machine learning model using the translated training data to predict words or phrases in the second language corresponding to the at least one intent;

link, responsive to generation of the translated workflow, the translated workflow to the autonomous chat agent, wherein the translated workflow configures the autonomous chat agent to receive messages in the second language and provide responses to the messages based on one or more intents of the messages identified via the translated machine learning model; and

update, responsive to receipt of the responses, the translated machine learning model based on feedback received from a user interface of the autonomous chat agent, the feedback indicative of an accuracy of the one or more intents reflected in the responses.

2 . The system of claim 1 , wherein the machine-readable instructions further cause the one or more processors to at least:

receive an override instruction for the at least one intent, the override instruction specifying a replacement value for at least one word or phrase in the translated training data; and

include the replacement value in the translated training data.

3 . The system of claim 1 , wherein the translation request further comprises one or more responses in the first language and the machine-readable instructions further cause the one or more processors to at least receive one or more translated responses from the machine translation service, the one or more translated responses having been translated from the one or more responses into the second language.

4 . The system of claim 1 , wherein the translation request is a first translation, the workflow is a first translated workflow, and the machine-readable instructions further cause the one or more processors to at least:

send a second translation request to the machine translation service, the second translation request comprising updated training data in the first language and the second translation request specifying the second language;

receive updated translated training data from the machine translation service, the updated translated training data having been translated from the updated training data into the second language;

create a second translated workflow to replace the first translated workflow, the second translated workflow comprising a second translated machine learning model and at least one translated intent;

add the updated translated training data to the second translated workflow; and

train the second translated machine learning model using the updated translated training data.

5 . The system of claim 4 , wherein the autonomous chat agent is a first autonomous chat agent and the machine-readable instructions further cause the one or more processors to at least replace the first autonomous chat agent with a second autonomous chat agent in the second language, the second autonomous chat agent being configured to use the second translated machine learning model to process user requests in the second language.

6 . The system of claim 1 , wherein the machine-readable instructions further cause the one or more processors to at least add a second translated intent to the translated machine learning model, the second translated intent being based at least in part on the second language or a geographic area associated with the translated machine learning model.

7 . A method to reduce processing cycles associated with execution of models used by autonomous chat agents, comprising:

receive, by one or more processors, from a service executed by the one or more processors, a request to update an autonomous chat agent, wherein the autonomous chat agent uses a workflow to process queries in a first language, wherein the workflow includes: (i) at least one intent, (ii), one or more responses, and (iii) training data, and (iv) a machine learning model configured to recognize the at least one intent in the first language and provide one or more responses based on the at least one intent

generating, by the one or more processors, responsive to receipt of the request, a translated workflow from the workflow, wherein the translated workflow corresponds to a second language and to generate the translated workflow, wherein generating the translated workflow from the workflow comprises:

sending, by the one or more processors, to a machine translation service, a translation request that specifies the second language, the translation request comprising the training data having words or phrases of the workflow in the first language, wherein the words or phrases correspond to the at least one intent of the workflow;

receiving, by the one or more processors, from the machine translation service responsive to the translation request, translated training data in the second language that is generated, by the machine translation service, from a translation of the training data in the first language, wherein the translated training data indicates words or phrases in the second language that relate to the at least one intent of the workflow;

and

training, by the one or more processors, a translated machine learning model using the translated training data to predict words or phrases in the second language corresponding to the at least one intent;

linking, by the one or more processors, responsive to generation of the translated workflow, the translated workflow to the autonomous chat agent, wherein the translated workflow configures the autonomous chat agent to receive messages in the second language and provide responses to the messages based on one or more intents of the messages identified via the translated machine learning model; and

update, by the one or more processors, responsive to receipt of the responses, the translated machine learning model based on feedback received from a user interface of the autonomous chat agent, the feedback indicative of an accuracy of the one or more intents reflected in the responses.

8 . The method of claim 7 , further comprising

receiving, by the one or more processors, an override instruction for the at least one intent, the override instruction specifying a replacement value for at least one word or phrase in the translated training data; and

including, by the one or more processors, the replacement value in the translated training data.

9 . The method of claim 7 , wherein the translation request further comprises one or more responses in the first language and the method further comprises receiving, by the one or more processors, one or more translated responses from the machine translation service, the one or more translated responses having been translated from the one or more responses into the second language.

10 . The method of claim 7 , wherein the translation request is a first translation, the workflow is a first translated workflow, and the method further comprises:

sending, by the one or more processors, a second translation request to the machine translation service, the second translation request comprising updated training data in the first language and the second translation request specifying the second language;

receiving, by the one or more processors, updated translated training data from the machine translation service, the updated translated training data having been translated from the updated training data into the second language;

creating, by the one or more processors, a second translated workflow to replace the first translated workflow, the second translated workflow comprising a second translated machine learning model and at least one translated intent;

adding, by the one or more processors, the updated translated training data to the second translated workflow; and

training, by the one or more processors, the second translated machine learning model using the updated translated training data.

11 . The method of claim 10 , wherein the autonomous chat agent is a first autonomous chat agent and the method further comprises replacing, by the one or more processors, the first autonomous chat agent with a second autonomous chat agent in the second language, the second autonomous chat agent being configured to use the second translated machine learning model to process user requests in the second language.

12 . The method of claim 7 , further comprising adding a second translated intent to the translated workflow, the second translated intent being based at least in part on the second language or a geographic area associated with the translated machine learning model.

13 . A non-transitory, computer-readable medium, comprising machine-readable instructions that, when executed by one or more processors, cause the one or more processors to at least:

receive, from a service executed by the one or more processors, a request to update an autonomous chat agent, wherein the autonomous chat agent uses a workflow to process queries in a first language, wherein the workflow includes: (i) at least one intent, (ii), one or more responses, and (iii) training data, and (iv) a machine learning model configured to recognize the at least one intent in the first language and provide one or more responses based on the at least one intent;

generate, responsive to receipt of the request, a translated workflow from the workflow, wherein the translated workflow corresponds to a second language and to generate the translated workflow, the one or more processors:

send, to a machine translation service, a translation request that specifies the second language, the translation request comprising the training data having words or phrases of the workflow in the first language, wherein the words or phrases correspond to the at least one intent of the workflow;

receive, from the machine translation service responsive to the translation request, translated training data in the second language that is generated, by the machine translation service, from a translation of the training data in the first language, wherein the translated training data indicates words or phrases in the second language that relate to the at least one intent of the workflow;

and

train a translated machine learning model using the translated training data to predict words or phrases in the second language corresponding to the at least one intent;

link, responsive to generation of the translated workflow, the translated workflow to the autonomous chat agent, wherein the translated workflow configures the autonomous chat agent to receive messages in the second language and provide responses to the messages based on one or more intents of the messages identified via the translated machine learning model; and

update, responsive to receipt of the responses, the translated machine learning model based on feedback received from a user interface of the autonomous chat agent, the feedback indicative of an accuracy of the one or more intents reflected in the responses.

14 . The non-transitory, computer-readable medium of claim 13 , wherein the machine-readable instructions further cause the one or more processors to at least

receive an override instruction for the at least one intent, the override instruction specifying a replacement value for at least one word or phrase in the translated training data; and

include the replacement value in the translated training data.

15 . The non-transitory, computer-readable medium of claim 13 , wherein the translation request further comprises one or more responses in the first language and the machine-readable instructions further cause the one or more processors to at least receive one or more translated responses from the machine translation service, the one or more translated responses having been translated from the one or more responses into the second language.

16 . The non-transitory, computer-readable medium of claim 13 , wherein the translation request is a first translation, the workflow is a first translated workflow, and the machine-readable instructions further cause the one or more processors to at least:

send a second translation request to the machine translation service, the second translation request comprising updated training data in the first language and the second translation request specifying the second language;

receive updated translated training data from the machine translation service, the updated translated training data having been translated from the updated training data into the second language;

create a second translated workflow to replace the first translated workflow, the second translated workflow comprising a second translated machine learning model and at least one translated intent;

add the updated translated training data to the second translated workflow; and

train the second translated machine learning model using the updated translated training data.

17 . The non-transitory, computer-readable medium of claim 16 , wherein the autonomous chat agent is a first autonomous chat agent and the machine-readable instructions further cause the one or more processors to at least replace the first autonomous chat agent with a second autonomous chat agent in the second language, the second autonomous chat agent being configured to use the second translated machine learning model to process user requests in the second language.