Artificial Intelligence – Background & Applicability
Although Artificial Intelligence as a term got coined way back in 1956 by John McCarthy, the applicability of this technology remained fairly dormant till recent times. There were however applications developed based on “Expert Systems”, such as DENDRAL, XCON, MYCIN, etc, the real power can be leveraged now, thanks to incredible improvement in compute power & network. Resurgence of AI and Cognitive system has been so dramatic, that the question being debated is not “If” we should be using AI/Cognitive but “When”.
Most of the IT companies have embarked on the path of building credible solutions built on Artificial Intelligence and Cognitive computing to deliver applications, which are supplementing human efforts in execution of IT projects and support engagements. LTI has a highly competitive and credible AI/Cognitive platform called Mosaic, which has been implemented for multiple customers worldwide. The building blocks of AI/Cognitive systems include not just a powerful inference engine, but also availability of strong knowledge base.
Application Support and Artificial Intelligence/Cognitive systems
Being one of the largest JDE practices in the world, we thought of investing in building a Chatbot-based support framework, specifically to address JDE related tickets. We all know that most of the organizations worldwide are under cost pressure and are exploring ways and means to minimise support cost. However, there is a danger as well if the cost is reduced at the cost of support quality. Quality erosion can directly impact the business resulting in serious consequences of losing responsiveness, customer satisfaction and market share. Therefore, it is, critical to find ways to minimise the cost without impacting the quality of support services.
If we take a close look at support services, there are a few levers, which directly impact the cost aspect. Following figure 1 identifies few such levers.
LTI Mosaic Run Solution
When we looked at these cost levers, it is apparent that the use of Cognitive Solutions can be an effective response to the cost and quality challenge all CIOs are facing. The essential ingredient for building a Chatbot-based support system includes a strong knowledge base of “known errors”, which can get enhanced over a period of time, as the Bot is operationally used (also known as Machine Learning). By virtue of supporting over 300,000 users worldwide, we have a robust KEDB (Known Error Data Base) in place.
The Bot developed is functionally supporting:
- End user by deciphering the problem and providing resolution if the issue is already known or reported earlier
- Logging the ticket on behalf of the end user facing a problem
- Helping the support team consultant by assisting in resolving the ticket.
- Auto-healing with resolving the ticket by making system changes (Advanced version)
The system learns as is put to use, which is the very tenet of any Cognitive system. This has helped in making the system mature over a period of time.
The overall flow is as per Figure 2 below:
Based on our experience, just by introduction of Bots, the overall activities gets significantly automated with reduction of human intervention. Following process flow in Figure 3 depicts the savings in terms of manual intervention in resolving the tickets.
Going forward, we will see many such tasks, which have been performed by humans, will be taken up by Bots. IT support services perhaps are the early adapters of such systems to compliment human tasks to give efficiency and scale of operations. The system described above is perhaps the first step in the journey of use of AI/Cognitive systems, and more is yet to come.
We have realised immense benefit by use of this system in specific modules of JDE. The overall efficiency gain is as less as 30% in some cases, however, it is going up to almost 70% in some other cases.