Sensible Case Classification With OpenAI – DZone – Uplaza

This text will exhibit how Kumologica and OpenAI can help in growing an AI agent API that effectively classifies circumstances inside an enterprise utilizing consumer information, with out the necessity for buyer assist agent intervention. An excellent case administration answer ought to have the ability to robotically categorize and prioritize circumstances from numerous channels, determine high-priority circumstances, apply labels, and assign them to the suitable teams with out requiring brokers to manually carry out these duties.

There are a number of case administration merchandise at present accessible available on the market, resembling ServiceNow, JIRA, and Salesforce Case Administration. Whereas a few of these merchandise embody built-in options for ticket classification, others supply no such performance. Even for these with built-in options, there are limitations of their capability to combine with third-party techniques.

Use Case

On this state of affairs, an ABC enterprise handles each IT and non-IT points via an enterprise portal. Customers present key particulars, resembling the problem title, description, and the related division (infrastructure, networking, CRM, and many others.), the place the problem ought to be submitted. As soon as the consumer completes these particulars, the case is submitted and enters a queue for a assist agent to evaluate. The agent then retrieves the ticket from the queue, critiques the data, and manually creates a ticket throughout the case administration system. Under Fig 1 offers the illustration of the prevailing case creation course of in ABC enterprise.

Design

New Structure

The structure workforce has proposed a brand new design to deal with delays in ticket processing and improve the effectivity of the ticketing system. They recognized that a good portion of the delay is brought on by the guide course of the place assist brokers learn, perceive, and classify tickets into the right division throughout the case administration platform.

In accordance with the brand new design, there’ll now not be any guide intervention between the consumer’s case submission on the enterprise portal and the case administration platform. The enterprise portal might be instantly built-in with the case administration platform via an clever case classification AI agent service. This agent will robotically interpret the user-provided particulars, categorize the ticket to the suitable division, and assign the right labels, tags, and tasks throughout the case administration system.

Technical Design

Within the new design, an API-based AI agent service might be developed to combine the enterprise portal with the case administration platform. This service will settle for requests from the enterprise portal, extract the outline particulars, and go them to a man-made intelligence supplier for classification. The AI will then return the required information to create the case throughout the case administration platform. Kumologica might be used as the combination framework to construct the AI agent in a low code style, using the OpenAI node to attach with OpenAI for clever classification.

Prerequisite

1. Obtain and set up Kumologica designer.

npm i @kumologica/sdk 

3. Set up OpenAI node within the Kumologica challenge package deal.

 npm i @kumologica/kumologica-contrib-openai  

2. OpenAI account entry and token for accessing OpenAI platform.

Steps

Now we are going to begin the implementation of smartcase classifier service in Kumologica. For many who are new to Kumologica, you’ll be able to refer the next tutorial hyperlink to get began. 

1.  Open the Kumologica Designer, drag and drop the EventListner node from pallet to the canvas. Present the next config for the node.

Show Title : [POST] /case/create
Supplier : AWS
EventSource : Amazon API gateway
Verb : POST
URL : /case/create

Notice: The article is taking the belief that ABC enterprise is having AWS cloud infrastructure.

2. Drag and drop Logger and wire it to the EventListener node added in the 1st step. Present the next config for Logger.

Show Title : Log Entry
Degree : INFO
Message : 'Request recevied : ' & msg.payload
Log format : String

3. Add two set property node to the canvas and supply the next config. That is to extract the request information and to set the principles for classifying the case.

Set-Property 1
Show Title : Set Case Knowledge

Operation : Set
Goal : msg.description
Supply : msg.payload.description

Operation : Set
Goal : msg.title
Supply : msg.payload.title

Set-Property 2
Show Title : Set Case Knowledge

Operation : Set
Goal : msg.rule
Supply : "if the outline comprises something associated to cloud, AWS, Azure, Sharepoint then present the next JSON. label is dependent upon if its AWS , Azure, Sharepoint and many others.
Description sttribute in JSON would be the unique description from consumer. 
{"Department" : "Infra-support", "Project" : "INF", "Label" : "AWS", "Description" : "" }
if the outline comprises something associated to community then present the next JSON. 
Description sttribute in JSON would be the unique description from consumer. 
{"Department" : "Network-support", "Project" : "NET", "Label" : "Networking", "Description" : "" }"

4. Now add the OpenAI node from the pallet and supply the next config. Then wire the node to the set property node in step 3.

Show Title : OpenAI
Operation : Single Q&A
Mannequin : gpt-4o-mini
API Key : >
System : msg.rule
Consumer : msg.description

5. Lastly we might be ending the movement with the response in EventListener Finish node. 

Show Title : Success
Response : HTTP Response
Standing Code : 200
Payload : msg.payload

The ultimate movement will seem like the picture beneath.

Attempt It

The above movement may be invoked through the use of the beneath endpoint regionally.

Endpoint : http://localhost:1880/case/create
Methodology : POST
Content material-Sort : software/json
Physique : {
  "title" : "Issue in with account access",
  "description" : "Request for providing the access the AWS dynamoDB in non prod account"
}

You may be getting the beneath because the response. The response reveals that the AI has labeled the division, challenge, label primarily based on the consumer description.

{"Department" : "Infra-support", "Project" : "INF", "Label" : "AWS", "Description" : "Request for providing the access the AWS dynamoDB in non prod account"}

Conclusion

Whereas integrating Kumologica and OpenAI for case classification gives a promising answer to streamline case administration, there are additionally vital issues. The automation can tremendously scale back delays and enhance accuracy, however enterprises ought to fastidiously assess their present infrastructure and potential integration challenges. Moreover, monitoring AI accuracy and making certain correct oversight might be essential for sustaining service high quality. As with all AI implementation, ongoing optimization might be vital to make sure that the system adapts to evolving enterprise wants and altering case dynamics.