Accelerate customer care using AI and genAI

We have contact centers, back-office support centers where employees of big companies and providers such as telco companies, insurance companies, banks, retailers, companies from energy providers solve customer questions, complains, requests. They need to categorize and prioritize customer interactions.

The agents spend a lot of time reading emails, web inquiries, sms, chats, phone calls transcriptions until the right action is addressed. In the CRM the interactions with customer are represented with activity date records which contain the transcription of the customer request. The agents read them and set fields such as<<category>>, <<type>>, <<priority>>. After this entity is based on the content of the interaction assigned to respective team. For example for issues with internet go to the technical backoffice team, questions about bills or prices entity is assigned to invoicing and billing backoffice etc. The amount of digital communication between companies and customers is growing exponentially. For insurance companies are these peaks seasonal , so to hire more agents will be expensive and not efficient. Outside these peaks they will not be used. How to react to these demands appropriate ? Artificial intelligence and generative AI is the best answer to accelerate customer care and react to the growing demand. 

We will demonstrate a way to improve categorization and prioritization of customer interaction requests for customer care backoffice or contact centers.

See this diagram.

We have Customer Relationship Management solution integrated with the contact center. The interactions are comming from omnichannels such as email, sms, whatsapp, phonecalls, letters which are scanned. All these interactions are represented after receiving from Contact center and sent to CRM as activities. The activity carries information with the customer text, in the case of phonecalls , chats transcriptions . The need to be assigned to respective team for finalization or further processing. To assign correct team either automatic assignment based on categorization or manually, when agent needs to read and understand the customer request. We can accelarate it to either help agent to understand it using AI to summarize the information, We can also extract entities and based on entities define rules for automatic assignment. Other accelaration is to understand customer sentiment and give the interaction a priority. Introduced genAI we can ask question about the customer request and based on answers do the assignment and prioritization.

Let us demonstrate 3 ways how to improve categorization of incoming interaction using AI services

1.  Summarize text using AI
2.  Analyze sentiment for prioritization
3.  Language understanding for intent identification
4.  Generative AI to categorize interaction

  Summarize text using AI

  Let us have following text , it is a complain about not working internet connections. This is coming to contact center and crm as an email from customer.

To Mobile Star, ltd
The manager, Mobile Star, ltd, Internet department
Respected sir,
With due respect, I am writing this email to inform you that my internet connection is disturbed for the last two days in my apartment. I am facing trouble as I am having exams, and most of the work is done online, and for that, I need a proper internet connection.
I have already made a complaint but no one took any action upon it. I request you again to send someone from your office to check my connection. My internet is during rain very slow and sometimes has outage. 
I am located in Bratislava, Namestie Slobody 11. My customer number is 111222 and my contract number is sample-1234. The connection speed in the evenings is not as purchased.
I expect a quick response as satisfying your customers should be your first priority.
If you fix my problem, I am about to order also your digital TV subscription with sport channels package. I am also thinking about upgrading my mobile phone subscription and get more data and also cheaper roaming calls.
Is there also possibility to get new phone if I sign commitment for the next 2 years ?
Thanks,
Yours sincerely,
Jozef Novak,
Strecnianska 1,
85101 Bratislava
Slovak Republic

 This customer text from interaction will be by trigger or batch process extracted and sent to cloud AI service such as Azure Language, AWS Amazon Comprehend or OCI Language. The response from cloud AI service can be like :

 Extractive Summary

 This summarized information can be stored to especial field of the activity or sent to another processing into cloud AI service for intent search.

 Analyze sentiment for prioritization

We can set rules that negative reactions have higher priority in order not to lose customer or fix his issue. Negative sentiment also means the customer needs more attention. In this case we sent the above text taken from 

 Analyzed sentiment

Based of the outcome we can set the field priority in order to process customer request faster.

Language understanding for intent identification

We can have trained our model for conversational language understanding for certain intents. Each intent can represent category of the interaction. We can have 1:1 mapping between company team/department and intent. If more intents are identified as in this case the interaction can be forwarded to more teams. In this case we forward it to <<internet technical support>>, <<hw sale>> and <<tv subscription specialist>>.