Struggling to build a conversational AI support assistant that doesn’t disappoint?

Conversational AI requires two key technologies - natural language understanding (NLU) to understand the question and branching logic to organise the conversation flow. If the process being automated is simple enough to be performed by inexpert humans the logic can be relatively simple. If the process requires human expertise the logic quickly becomes impossibly complex. Frameworks like IBM Watson or Google DialogFlow combine rocket-science natural language (based on machine learning AI) with Soyuz-era logic networks of IF..THEN statements. The result, for anything but the most trivial business processes, is AI assistants that excite by appearing to understand questions but disappoint by failing to answer them.

eXvisory is a visual development framework for scalably and maintainably building deep logic support assistants. Using it is fun and a bit like solving a jigsaw puzzle. It can be used standalone or as a ‘skill builder’ plugin to other AI assistant frameworks.

Chatbot Decision Trees How hard can they be?

Shallow chatbots are boring Adding deep logic to Google DialogFlow

Do you need to be a programmer to build a chatbot?

How do you add deep logic (expertise) to a chatbot?