Experience the power of sales development automation for Customer Service

How to create outlook signature

[Music] welcome to the demo of a customer service fulfillment automation for banking using uipath and its strategic partner h2o with this demo you'll see how banks and others can reduce the time it takes to triage and help resolve incoming email based support requests from minutes to seconds banking customers often ask for help or raise complaints through email or web requests they also communicate with bank support staff through email until the request is successfully resolved as a result banks customer service teams have to process a great number of unstructured text based communications they also have to send data back and forth between the customer and their banking business applications these steps are often manual so they take a long time to resolve cause great dissatisfaction to customers and take hundreds of staff to process this demo shows how uipath and uipath business partner h2o use robotic process automation and natural language processing to automate a key step of the customer service experience the classification of incoming requests into specific banking product categories let's take a look at an example inbound email customer request triage automation the robot will check for new emails if it finds none the process ends if it finds a new email the robot will then classify it and route it appropriately then start all over again with UI paths drag-and-drop AI capability we are able to create a customer activity called classify email using HTTP the activity calls a machine learning model to classify emails based on the text within the body of the email here we use a natural language processing model from our business partner h2o the h2o model returns a decision for which support class to use and a confidence level on its decision because it's an activity it's reusable in other process flows the robot then checks whether the confidence level is high enough here we set confidence at 80% if confidence is lower than 80% the robot sends the email to an unclassified folder for a person to review if the confidence level is higher than 80% the robot sends the email to the folder assigned for that class here you can see a Microsoft Outlook email inbox with inbound requests emails that have not yet been classified once the automation is run it will loop through this folder and move the emails to the correct folder for follow-up the robot could be configured either as an unattended robot that sends lower confidence emails to an unclassified folder or as an attended robot that asks a person for where to route an email it can't classify it took about nine seconds to classify 17 emails in addition the robot has captured a log that shows how long it took for each email the destination the email was sent to and the confidence level for the decision taking a quick look at each of the destination folders we can see the emails that have been moved and classified requests for checking accounts credit cards mortgages and more when we open one of the emails that was classified and sent to the student loan folder you can see that the body of the email didn't mention the word student once despite this the model was able to accurately predict the email was about a student loan question and with a reasonable confidence level directed the email to the student loan folder with automated email classification banks can significantly improve customer service reduce their operational footprint and better handle seasonal variations of support requests for more information visit us online at uipath accom slash solutions slash industry slash banking or talk to your uipath or automation partner banking specialists