• describe relations in 5 lines
• preview on your machine
• deploy to any Spark cluster
• scale up
• deep web => data product, in realtime

It's a Deploy-&-Forget thing

• mimic Spark ML pipeline stages
• takes 5 mintues to assimilate a web service, 30 seconds if someone has already done so
• become permanently effective once shared online: will be tested daily by our servers and adapt to your data provider
• now get back to data science and forget it ever exists

In Soviet Russia, your AI trains you!

• predictive AI
• | name doesn't ring a bell?
• | Google it!
• | Keyword/Entity extraction
• | Still can't predict credibly?
• | post to crowdsourcing platform!
• V get answer & retrain
• proactive AI