Lior grinberg1
Tera Contributor

So, I got a requirement from a customer to find a way to read attachments content of incident in order to find if they hold finance data like invoices, bank statements and then associate the incident to relevant assignment group. When I analyzed the request, I learned that these are my challenges:

  • Attachment name – in most cases meant nothing,
  • Attachments type – there were many of those, including PDF, MS Office documents, MS Excel, and Images
  • Attachment Language - the attachments were in different languages

  

While analyzing the problem, I realized that data in the attachments could help me much more than I expected, and not only for arranging invoices. Almost all incidents/cases opened by the users had a file attached to it - a screenshot that points to the error message on the screen, or a file that holds important details.

A simple example of such an incident is a user that reports a malfunction with his application and adds a screenshot with the error message that he sees. If we better understood that image, we could propose the user a Knowledge Article related to the problem, or auto-assign the Incident to the correct support group.

Let’s take a look at an example, where a user reports a problem in an SAP screen and he uses the following image as an attachment:

find_real_file.png

Based on the screenshot title - “Create Purchase Order” - we can assign the case to the “SAP Procurement” support group and enter the correct category and subcategory automatically.

The solution I have created is an application that uses your mid server; it downloads only the relevant attachments and retrieves the content of the attachments by means of OCR for images, or by accessing the content of other types of attachments and sending back the results to ServiceNow.

I wanted to share my solution with this great community to get feedback and possibly more ideas on how to expand the application, what kind of useful automation can be added to it after getting the content of the data, and more.

You are most welcome to share your thoughts!

Thanks,

 Lior Grinberg

Version history
Last update:
‎10-22-2020 12:04 AM
Updated by: