Scene 1

[Audio] The objective of BDE is to respond to members realtime and well within the 8 hour SLA. The solution also streamlines responses and moves away from the uses of multiple mailboxes. With the use of Artificial Intelligence, we are able to categorize the email request and follow the BU process to action the request.
Anti selection of queries, i.e. the process works the oldest to newest queries
Reduce Duplicate queries by actioning the members request as soon as the email is received. Because the solution refers to specific benches, dedicated teams action the queries once received in the workbench. As the solution matures we will further define the assignment of queries that require human intervention, to promote the blended model. No-reply e-mails, out of office responses e.g will be identified as SPAM and closed without being reassigned to the BU. 

Scene 2

Problem​ Resolution​ Not Real-time ​ ​ SLA 8  Hr ​ ​ 800+ mailbox ​ ​ No accurate categorization ​ ​ No workflow – manual selection​ ​ No Escalation unless set up​ ​ No Duplicate control​ ​ No-reply email addresses are received​ ​

[Audio] This the first Cloud based solution that has been successfully implemented into a production environment! It is also our first Hyper-automated solution which means that a combination of different technologies was used. High level the technologies used were the Read mail engine which is an existing e-mail validation engine. We then introduced NLP and logic app and mail 365 to support the receipt and interpretation of the mail. Once interpreted categorization takes place and the robot is introduced into the process to execute the mail request within the core application. In this instance that would be Nexus. In Nexus we use the same set-up as any other BU, we have a workbench which contains tasks that will be actioned as per the standard human user process. NOTES- Not for Recording. Office 365 is a Software as a service or SaaS solution that provides multi-tenant options, for example Power Automate workflows and exchange online for e-mails. Office 365 also includes Power BI and provides build-in security that helps to provide a productive and secure remote work environment. It is a platform offered and supported by Microsoft as a cloud service. Azure is complementary to the Office 365 infrastructure. Azure allows us to run our platform as a service where custom applications can be created and published. It also provides a cloud service environment complete with load balancing and no on-premises servers are required. The BDE solution includes the use of existing code within the Readmail engine, Natural language processing, the logic app used for classification of emails, and RPA Blue prism for robotic process automation. 

Scene 3

A picture containing text, sign, device, gauge

Description automatically generated

Using AI such as NLP/Logic App in conjunction with RPA  BluePrism  creates intelligent machines that simulates smart behavior and support decision making with little to no human interference

[Audio] Of all the technologies used to deliver the solution the Natural Language Processing technology links the receipt of the email to the process that needs to be executed. How does NLP do this? Natural language processing ( NLP) is a branch of artificial intelligence that helps computers understand and interpret human language
Natural language processing will be used to interpret and analyze unstructured data received in e-mail queries. Sentiment ( words and phrases in an email) will be fed into NLP. Together with Logic App, the technology will prompt the digital user to complete the request. 

Scene 4

Natural language processing (NLP) is a branch of  artificial intelligence   that helps computers understand, interpret and manipulate human language

It fills the gap between human communication and computer understanding.

Today’s machines can analyze more language-based data than humans, without fatigue and in a consistent, unbiased way. Considering the staggering amount of unstructured data that’s generated every day automation will be critical to fully analyze text and speech data efficiently.

[Audio] So how does this end to end process work?
The email is received from the member to the scheme mailbox The Readme engine validates the email received to validate against the membership details in Nexus. A Noc query is then created in Nexus-on-call and the received email saved as member correspondence NLP and Logic will classify the emails into the relevant folders within the mailbox based on the sentiment in the subject lines of emails received. The digital worker will then action each request based on the classification done by NLP/Logic and reply using scheme specific email addresses Three different actions can then be taken : STP that will complete the query, or An e-mail response request to the member requesting more information to assist with the completion of a query I need my human counterpart to assist me further, so I reassign the query. 

Scene 5

Email received in
Scheme specific
mailbox
Readmail
validations
performed on
email in relation to
NEXUS
NOC Query is
created in NEXUS
NEXUS
In Scope queries
are closed and
COMPLETE
NLP/Logic listener
will classify the
email based on
email subject
More info
requested from
member. Query
closed and
COMPLETE
Out of scope is
reassigned as
BUSINESS
EXCEPTION
Human intervention
required. Query
reassigned and
COMPLETE

[Audio] To reiterate some of the benefits that we will start seeing with implementation of the solution is:
Move email to realtime
Reduce the response times
Reduce transaction times
Reduce repeat queries
Improve CSAT score
Deliver Service in the Moment. 

Scene 6

[Audio] Some of the value adds that we have introduced during the soft go live with Polmed is: With additional SPAM filter we're able to do this by using words and phrases to identify SPAM in order to close these queries instead of reassigning these to the BU
We are reassigning to specific workbenches as defined by business that will contribute to the realization of the overall vision in respect of the blended model
This is where a Bonitas member may send a query on behalf of a Polmed member. Previously the Readmail validation would have indexed the email against the Bonitas member, which meant that a Bonitas membership number would appear in a Polmed workbench as a member query. Usually when this happened a manager had to reassign the query to be actioned. Through the solution we were able to update this validation even though the Bonitas membership number would still appear in the Polmed workbench, a human or digital user will be able to action the query without manager or TL intervention
The correspondence component of the process will continue to promote the use of omni channels e.g. Whatsapp Sentiment updates are daily, that means we are training our technology daily to improve scope of interpretation. 

Scene 7

Correct routing of queries – supports blended model

Locked membership numbers can be actioned

Encourage the use of  omni -channels e.g. Web and WhatsApp

NLP/Logic app –Daily sentiment training

Real-time Bidirectional Mail Solution

The objective of BDE is to respond to members realtime and well within the 8 hour SLA

The solution also streamlines responses and  moves away from the uses of multiple mailboxes

With the use of Artificial Intelligence, we are able to categorize the email request and follow the BU process to action the request.

Anti selection of queries, in other words the process works the oldest to newest queries

Reduce Duplicate queries by actioning the members request as soon as the email is received.

Because the solution refers to specific benches, dedicated teams action the queries once received in the workbench

As the solution matures we will further define the assignment of queries that require human intervention, to promote the blended model.

No-reply e-mails, out of office responses e.g will be identified as SPAM and closed without being reassigned to the BU

avatar

This the first Cloud based solution that has been successfully implemented into a production environment

It is also our first Hyper-automated solution which means that a combination of different technologies was used

High level the technologies used were the Read mail engine which is an existing e-mail validation engine.  We then introduced NLP and logic app and mail 365 to support the receipt and interpretation of the mail.  Once interpreted categorization takes place and the robot is introduced into the process to execute the mail request within the core application

In Nexus we use the same set-up as any other BU, we have a workbench which contains tasks that will be actioned as per the standard human user process.

Office 365  is a Software as a service or SaaS solution that provides multi-tenant options, for example Power Automate workflows and exchange online for e-mails

Office 365 also includes Power BI and provides build-in security that helps to provide a productive and secure remote work environment

It is a platform offered and supported by Microsoft as a cloud service.  Azure is complementary to the Office 365 infrastructure

Azure allows us to run our platform as a service where custom applications can be created and published

It also provides a cloud service environment complete with load balancing and no on-premises servers are required.

The BDE solution includes the use of existing code within the Readmail engine, Natural language processing, the logic app used for classification of emails, and RPA Blue prism for robotic process automation

Of all the technologies used to deliver the solution the Natural Language Processing technology links the receipt of the email to the process that needs to be executed

Natural language processing (NLP) is a branch of artificial intelligence that helps computers understand and interpret human language

Natural language processing will be used to interpret and analyze unstructured data received in e-mail queries

Sentiment (words and phrases in an email) will be fed into NLP. Together with Logic App, the technology will prompt the digital user to complete the request

slide4-hotlinkundefined

slide4-hotlink4

 Natural language processing (NLP) is a branch of  artificial intelligence   that helps computers understand, interpret and manipulate human language

slide4-hotlink31

So how does this end to end process work?

The email is received from the member to the scheme mailbox

The Readme engine validates the email received to validate against the membership details in Nexus

A Noc query is then created in Nexus-on-call and the received email saved as member correspondence

NLP and Logic will classify the emails into the relevant folders within the mailbox based on the sentiment in the subject lines of emails received

The digital worker will then action each request based on the classification done by NLP/Logic and reply using scheme specific email addresses

Three different actions can then be taken :

An e-mail response request to the member requesting more information to assist with the completion of a query

I need my human counterpart to assist me further, so I reassign the query

To reiterate some of the benefits that we will start seeing with implementation of the solution is:

Some of the value adds that we have introduced during the soft go live with Polmed is:

With additional SPAM filter we're able to do this by using words and phrases to identify SPAM in order to close these queries instead of reassigning these to the BU

We are reassigning to specific workbenches as defined by business that will contribute to the realization of the overall vision in respect of the blended model

This is where a Bonitas member may send a query on behalf of a Polmed member

Previously the Readmail validation would have indexed the email against the Bonitas member, which meant that a Bonitas membership number would appear in a Polmed workbench as a member query

Usually when this happened a manager had to reassign the query to be actioned

Through the solution we were able to update this validation even though the Bonitas membership number would still appear in the Polmed workbench, a human or digital user will be able to action the query without manager or TL intervention

The correspondence component of the process will continue to promote the use of omni channels for example Whatsapp

Sentiment updates are daily, that means we are training our technology daily to improve scope of interpretation