Is there a way to read an email body and create tasks based on certain keywords ?

Certified Associate Developer

We have got an email box ( for example : Customercare@xyzcompnay.com). There are thousands of emails here related to customer issues like 

Email 1 : "internet is very slow" 
Email 2 : "have a query regarding refund" .etc.

Email X: "Would like to change my personal details".

I would like to create a process which reads the above emails and creates tasks based on keywords.  This process has to create as task for Finance team with Email2 as it is related to Refund.

And it has to create another task for IT team with Email1 data as it is related to Internet and so on another task to HR Team with Email X.

I assume it is more of a AI ( Machine learning ) work.  Is there any way to do this in Appian ?

Thanks

  Discussion posts and replies are publicly visible

Parents
  • Yes, this is certainly possible.  Once you are configured to receive email to start a process, you will essentially create an algorithm that takes the subject / body of the email, parses it for key words and decides which outcome is best.  The key will be in tuning your algorithm and/or applying machine learning based on process outcomes, etc.

    I had a process a few years back that handled user support for a fairly high use application, requests came in via email.  The system would automatically handle one type of request, specifically archiving processes.  When keywords for this archival request were found in the body, along with the request number, etc, the system would archive the request and submit a follow-up task to the user for confirmation - giving them one click to confirm they were all set or one for 'that was not my issue', which unarchived and logged as a a false positive in the algorithm.  All OOTB Appian.

Reply
  • Yes, this is certainly possible.  Once you are configured to receive email to start a process, you will essentially create an algorithm that takes the subject / body of the email, parses it for key words and decides which outcome is best.  The key will be in tuning your algorithm and/or applying machine learning based on process outcomes, etc.

    I had a process a few years back that handled user support for a fairly high use application, requests came in via email.  The system would automatically handle one type of request, specifically archiving processes.  When keywords for this archival request were found in the body, along with the request number, etc, the system would archive the request and submit a follow-up task to the user for confirmation - giving them one click to confirm they were all set or one for 'that was not my issue', which unarchived and logged as a a false positive in the algorithm.  All OOTB Appian.

Children
No Data