How AI Drives Your Incident Management and Incident Creation Processes #SAP User Group Webcast Summary

From the abstract:
What is AI driven support and how does this improve your incident management and incident creation process?

SAP’s incident management process has been considerably optimized and simplified by AI and machine learning technologies. Today, solutions like Incident Solution Matching or the component predictor help customers in an optimized and simplified way to find answers to their questions faster and easier. Also the creation and management of incidents has been improved considerably. For example, as a part of our machine learning–empowered support processes, solutions are automatically proposed within the incident creation form within SAP ONE Support Launchpad so customers can find relevant SAP Notes and SAP Knowledge Base Articles quickly without searching manually.

This webinar will provide you an overview on the AI driven processes and how the customer support experience has been improved by implementing AI technology and machine learning. Join the session and discuss your questions with our AI experts.  Source: SAP

Vision of SAP Support

Create new element, AI driven support

Source: SAP

Self service prevention, identify the issues
Real time interaction, speed up time to issue
create a native support experience, seamlessly integrate intelligence
Learn from previous incidents

AI Powered Support Services

Figure 3: Source: SAP
Today incident solution matching; see it on launchpad

Figure 4: Source: SAP
Component area prediction – SAP suggestion – coming from categorization service

Figure 5: Source: SAP
Expert area service – suggests the most suitable areas to be related to topic

Figure 6: Source: SAP
Where available

Plan to integrate in SAP Communities, and SAP help

Figure 7: Source: SAP
This was explained in a previous webinar (I missed)

ISM service – pre processing, tokenize the information, historical incidents and the solutions

Frequency and algorithm – create most prominent solution and documents; compared with the input

Signals are the features, labels, key part of the document collected, will identify the best candidates from first cut

For each signal, give a weight, process signals based on weight, final ranking

Another signal is KBA, notes, and recency

Attachment rate is another signal

Create training model, and retrain model in different ways – using incident resolved, solution with customer, and interaction with customer

Figure 8: Source: SAP
Running on a platform
Services run on an API
Using MongoDB
Tensorflow
Kubernates
Run on cloud internally
Cubeflow for the pipeline

Figure 9: Source: SAP
Last benchmark, 54.7 percent of cases in top 10, offering the right solution
repositories today KBA,

Future, Help, Wikis, blogs and more

What’s next?

Figure 10: Source: SAP
Interactive ISM
Interactive mockup is on the right; enter terms, once refined, list is refreshed, new solutions suggested
Based on the information in the incident
Filters are shown

Create a conversation with customer
Results

Solution navigation service, giving customer the possibility to navigate to similar notes
Will be part of the incident creation in the launchpad

Question & Answer

Q: What is the timeframe for future improvements?
A: First version of interactive ISM is in March
Will be a continuous improvement

Q: In terms of the SAP Community will ISM propose answers or tag matching?
A: Propose answers – still open to different places
Tag matching – discussed

Q: How will ISM be incorporated in SAP Help?
A: still under discussion…search, interest

What do you think?

I am looking forward to the possible tagging help on the SAP community site; imagine you ask your question and it populates the tags for you.