AI/ML has been the buzzword that has dominated the tech industry for some time now. How are contact centres using these buzzwords to improve agent's efficiency and customer satisfaction?
We are laser-focused on AI/ML features which can solve service/contact centre's day to day problems, improve productivity, acheive target SLAs...
What is the current state of many contact centers?
Many Oracle B2C Service Center customers use product/category/disposition fields for routing the incidents to the agents with the expertise to improve resolution time. However, the incident created might not always have the fields filled in when raised via channels like Customer Portal (Self Service) Webform, Service Emails, and Live Chat.
And Some channels like service email, may not even have the provision to collect values for such fields, that can be used to automate the routing. Some of the B2C customers use keywords for automating such cases, but the accuracy and efficiency tend to be very low.
OK, what we have released in our 23A release?
Our Incident/Service Request Classifier accelerator will be helping our B2C Service Center customers to take a step forward in the direction of AI/ML innovations. As you might have guessed, the solution classifies the incidents, based on “incident subject & body” created in B2C service to improve incident routing to accurate queues or agents.
With our 23A accelerator, we are providing a feature, where the customers can integrate B2C Service with OCI Data Science services, and use the ML capability to predict products, categories, or disposition fields.
It leverages native B2C Service CPM scripts and extension features to integrate the Incident Classifier capabilities of the ML model for the incidents created via webform (Customer Portal - Ask a Question) and service email.
Now, what are Accelerators?
Accelerators are the sample code, and integration stack configuration released along with documentation to implement for the customers to solve their problems by integrating to other services using B2C Service's capabilities under UPL License. (For more details on UPL, see- https://oss.oracle.com/licenses/upl/). This sample code, sample AI/ML model can be customized as per the customer’s unique business use cases and solve the issues specific to them, and hence customers can use this as it is or as a starting point. This sample code is to demonstrate the integration capabilities of B2C Service with other products.
What does Incident Classifier do?
It predicts the product, category, and/or disposition field(s) of new incidents created in B2C Service based on the text of the Subject and Body fields. The prediction is made using the OCI data science ML capability. It works on a deployed ML model which is trained using past data of the first thread of closed incidents. The model can be re-trained as frequently as the customer wishes. However, for a model to be effective, the model initially needs constant training and this is made configurable in the solution provide, so the frequency of re-training can be adjusted as per the dataset and iteration.
What does Incident Classifier architecture look like?
Any custom trained AI/ML model needs three phases, i.e. Ingestion, Build and Deploy phases. And in this accelerator, it uses OCI job, object storage, Model artifact and Oracle Fn to achieve the same. Once the model is deployed, the service is exposed to public using API gateway. These services are invoked from B2C Service platform via CPM/extensions. The architecture which depicts this flow is shown below,
Business values provided by Incident Classifier feature?
Okay, got it, how can we access this AI/ML accelerator?
Well, there are two options for accessing the accelerators
Do we have any Prerequisites? - Yes:
OCI Tenancy: Please reach out to the Oracle sales team or Oracle support team to set up the OCI tenancy.
Tips & Considerations:
Questions or feedback?
We are really excited for you, to try this feature.
More Documentation for Reference
OCI Documentation: https://docs.oracle.com/en-us/iaas/Content/GSG/Concepts/baremetalintro.htm