Hello, how can we help you?

Recent Searches: Workspace - 3.1 - Workflow - 3.1 - Training - 3.1 - Request - 3.1

Automating workflows with AI

API icon with a blue background, which leads to the initial stage of configuring the API.
Robot icon with a blue background, which leads to the stage of using AI in SoftExpert Suite.

Prerequisites

 

Introduction

In the BPM solution (composed of the SoftExpert Workflow, SoftExpert Form, and SoftExpert Process components), Artificial Intelligence (AI) is used to optimize and transform business processes through the automatic execution of activities. 

This article will demonstrate the use of AI in the BPM solution to automate workflows.
 

Configuring commands and web services

Before using workflow automation, it is necessary to configure a command and the web service in the system activity that will receive the automation. See an example below:

1. Access the Process (PM063) menu.

2. Search for the desired process.

3. With the process selected, click on the Open flowchart button.

4. In the process flowchart, double-click on the system activity in which AI will be applied.

5. In the General data tab, in Integrations, select Web Service

6. Click on Web Service configuration and configure the SoftExpert Ask AI data source.

7. In the Formula (Before execution) field, click on Formula editor and enter the command that the AI will receive.

Command example

//generate prompt

prompt = "You are an incident specialist who is responsible for analyzing an incident description and decide if is or not is an incident. "
prompt = prompt +" Tag <incident> contain the incident description."
prompt = prompt +" <incident>"
prompt = prompt +" "+ FIELDVALUE('ITIL';'txincident')
prompt = prompt + " <\\incident>"
prompt = prompt +" The answer is only YES or NO."
prompt = prompt +" The response should be a single text (YES or NO) that represents if is or not is an incident."
prompt = prompt +" Do not return an explanation of why you selected it."


RETURN DEFINESTABLEFIELD('ITIL';'txprompttriage';prompt)
 
 

8. After inserting the formula, click on Save and exit.

9. Having configured the command and the web service, click again on Save and exit. Thus, the automation will be ready to be executed.

 

Operation of the workflow automation in practice

To understand how the workflow automation works, consider the following context: an organization has an incident management process, which brings together all the activities necessary to identify incidents. 

See below examples of how AI can be used in each step of the incident management process:

Process step 

Use of AI
Incident recording
  • The user records the incident in the form
  • When the process moves to the next step, the AI ​​will automatically triage it and decide whether or not it is an incident.
  • The IA can also be used in the form filled out by the user to help in the incident description.
 Triage
  • The AI performs the initial triage, identifying whether the record is an incident or not.
 Categorization
  • The AI categorizes the incident to send it to the most suitable team. If it is not considered an incident, the process is closed. 
  • It is possible to see the path followed by the process by accessing the flowchart through the execution activity.
Sending to the responsible team
  • The IA sends the incident, through My tasks > Activity execution, to the responsible team.
  • The responsible team must complete a resolution for the incident and, if necessary, can use AI to improve the description of the resolution before sending the process to the next step.
Impact analysis
  • The AI ​​generates an impact analysis according to severity and decides whether it will be necessary to create an action plan, provide feedback, or both.
Incident investigation and diagnosis
  • If the AI ​​identifies that the incident is high-impact, an in-depth investigation with an action plan will be required.
  • It is possible to use AI to generate the actions of the action plan.
User feedback
  • Uses AI to generate sentiment analysis and insert it into the form.
  • The sentiment analysis uses the SoftExpert Sentiment Analysis data source, which has a pre-configured command to analyze texts and identify the predominant sentiment, classifying it as positive, negative, or neutral. In addition to the classification, a numeric score is provided for each sentiment, allowing for a more detailed analysis.

The following video illustrates the information in the table:

 

Conclusion

The use of Artificial Intelligence (AI) to automate workflows not only facilitates triage, categorization, and analysis, but also contributes to more agile management, minimizing errors and increasing the accuracy of decisions made throughout the steps of the process.


Was this article helpful?