I wrote about the complexities involved in implementing NLP (Natural Language Processing), wherein I discussed the technical aspects of it. Today we will talk about the business perspective of why a company would want to implement NLP.
Sometime back, I got a call from a major financial service provider company asking if I would be interested in taking up a full-time position with them which involved AI and Machine Learning. A significant portion of the work would involve handling tech. related to omnichannel customer service, which included emails, phone IVRs, chatbots, and customer feedback and ratings through their website. They said they had all the top-notch AI platforms and tools, and they would like an AI architect to put it all together and cook a solution for them.
I asked them what are their priorities, and why would they want to do so. Although most of the priorities are the same for many companies, I wanted to check, and their priorities were not surprising:
- Too much of billing is happening from customer service staff attending customer calls
- It is tough to have a single unified dashboard of customer feedback and the list of current issues
- They were struggling with NLP, and it seemed like an uncontrollable beast!
Although their priorities seemed straight-forward, I felt that they had not done their homework of why the billing from staff was high and how to reduce it. It seemed like they were in a hurry that NLP and the modern tech. would solve the problem with humans out of the picture. I didn’t think so, but I didn’t tell that to them because they were not in a mood to listen. I politely refused their offer and asked them to contact me for any point solutions that would solve a specific technical issue instead of a complete overhaul to tech.
Many companies are in a hurry, fancied by the glitter of new tech. As I wrote before, NLP is complex and needs to be carefully implemented. Buying an NLP ‘package’ and thinking of integrating it to your customer service will not be the right approach, because context matters. If implementing itself is tough, imagine how complex it would be to test!
Software Quality is not just about metrics, but mostly about getting the objectives, priorities, and requirements right, right from the beginning of the project. If something does not make sense, you got to let the customer know, so that they can avoid the later heartburn and spend of so much of money.
Subscribe to my YouTube channel for discussions on various aspects of technology and how Software Testing can help. There are discussions about Software Quality too. Here is the link:
What are your priorities as a business to implement AI and Machine Learning in your companies?
Contact me for a discussion!