[abtinf]

This is a great simple example of building a bot and could be a nice starting point for a lot of of projects.

If you are interested, there are more advanced techniques as well. For example, https://github.com/watson-developer-cloud/conversational-age... shows a conversational agent that combines a natural language classifier with a dialog engine. The cool thing about using NLC is that you can infer user intent from ambiguous inputs to make a more robust bot.

It will be interesting to see if all this new interest in bots represents the coming of useful bots or if it's just a fad. I love using bots in specific, tool-like ways, like getting github updates. But seeing someone build up a bot that can really tackle natural language to do useful things in a predictable manner would be amazing and could change how we interact with computers in general.

Disclosure: I am a Watson Developer Cloud Evangelist

[Bjartr]

Does this system use the surrounding context of the conversation when doing its classification, or is each statement processed independently of what came before it?