Serverless Facebook Messenger Bot

Overview

Listen — receive a request object from Facebook Messenger, and extract the sender’s userid, text, and attachments.
Remember — check records for any past conversations with the same user, and retrieve any previously stored context. If this is the first time the user has spoken to our bot, create a new record with their id and empty context.
Pre-process — handle possible input types; images, audio, and location pins. Classify images using visual recognition, transcribe audio using speech-to-text, extract coordinates from location pins, and finally, send the input to Watson Conversation and receive its response.
Process — resolve any pending actions as indicated by Watson Conversation, fetch data from Cloudant if needed, and append to output.
Memorize — update the user’s record with the new context.
Respond — format output and send it back to Facebook Messenger.
Webhook — to handle Facebook authenticatoin requests.