Rasa

Described as an open-source conversational AI package, Rasa seems to have it all. It centers on formulating dream virtual assistance based on developing truly contextual machine learning. On this basis & deploying the following tools, Rasa clearly has a wonderful array of features:

• Natural language understanding (NLU) to categorise queries, reply & acquire pertinent info accordingly
• Core dialogue engine management to augment virtual reality & the human touch
• An extension of this is voice applications for IVR, conversational search, mobile apps, home controlled device automation & smart speakers
• Interactive updated research-based learning & view/annotate correspondence to improve the chatbot’s understanding ability to handle queries
• Curate multiple versions with the option to revert the changes or use an earlier program
• Ready to apply & implement via the in-house system or on any cloud-based platform

Rasa is winning so far but what about the shortcomings it experiences? Well, firstly whilst browsing their website to write this review, I noticed a simple but important flaw. Trying to access their section re ‘multiple integrations & channel options’ returns with a broken or non-existent URL. Not a great first impression in real life (no pun intended). Secondly, the main issue surrounds its bot-building capabilities, which regardless of their assurance, still remain an expert’s choice.

Therefore, Rasa explores unchartered territory when exploring the scope of chatbot use. Users are able to literally train a bot to learn how to deal with human inquiries & learn how to improve upon its responses. This is great, however, may hinder the not so technically adept users from fully utilising this concept. Hence, it may not be possible to fully reap the benefits of what is otherwise a comprehensive automated solution for customer service. It’s a shame really, owing to the promising prospects it can bring.