Chat-bot development

MultiMed Engineers has started experimenting the application of chat-bot technology to several business cases, proposed by its customers. In particular, it has developed two “proof of concept” MVPs, that explore the following business cases:

  • NLP-based “query language” for a CRM database, to enable sales agents to quickly retrieve data about customers “on the spot”, through an intuitive, mobile chat-bot. This work includes the identification of “intents”, to translate relevant query verbs to/from natural language, and the extraction of “entities”, to denote relevant database objectsTinyCRM2
  • Conversational application to support a business-to-business “classifieds” system, to search for and offer opportunities for partnership, collaborations, investments and the like, among Italian SMEs. This work included the design of “stories” to represent patterns of conversation in the chat-bot (see screencast below). A production instance has been active for a test period at the Officine Italiane Innovazione Facebook page (visible at work in the video below).

 

 

Both cases have been developed on the Wit.ai NLP conversational platform, integrated into the Facebook’s Messenger instant messaging app.

The following Python frameworks have been used in development:

 

Example of a code snippet

Example of a code snippet

 

chatbot_tdd

Example of a TDD scenario