Note: This project is still in its infancy and is subject to rapid/breaking changes.

The goal of this project is to provide a simple but powerful way of designing application-specific chatbots. By application-specific I mean the conversations have a particular context and goal, such as gathering information from a consumer or providing answers to domain-specific questions.

This project sprung out of a disappointment with the flexibility allowed by today's popular tools such as DialogFlow and Amazon Lex. The aim is to go beyond the features they offer in terms of controlling the flow of the conversation. At the same time, the project aims to provide the flexibility to rely on parts of those services. In particular, you can roll your own NLU (see note below) and add text-to-speech capabilities as needed with whatever service you want.

Note: currently only LUIS is supported. I plan to add support for DialogFlow, Lex, and Rasa at some point, but LUIS exhibits solid performance and has the ability to flag multiple intents at once.

Documentation is quite limited at this time. Stay tuned.

MIT