Talk in English - US at Dutch PHP Conference 2018
View Slides: https://speakerdeck.com/gbtekkie/dpc18-a-lamp-stack-professionals-journey-to-artificial-intelligence-meet-johnny-the-deployer
Checkout the code: https://github.com/tekkie/johnny-deployer-demo
Short URL: https://joind.in/talk/e45b2
(QR-Code (opens in new window))
Artificial intelligence products are all the rage these days. AI is not scary, and there are a lot of awesome things you can do with it right from PHP. How do we, as LAMP stack professionals, ensure we embark this wonderful journey? I’ll show you how we built, with PHP, a devops chatbot that enables the release manager skip the confusing discussions with the tekkies and master the release process. It’s not just a simple bot, because we added from day 1 natural language processing smartness to it, so that our user base is extended to the project managers, business analysts, and customer engagement teammates. Johnny started as a devops friend for one of our customers, being extremely helpful in our day-to-day tasks. It later evolved to a tech lead/scrum master role, and even reminds us to do our timesheets. Here are some discussions you might have with Johnny: “What tickets are ready for QA?” “What version of the API is installed in production?” “Are there unused branches for the website?” "Has test coverage decrease in the last week for the API?" "What is the development team on the ETL project?" "Add Mary as a developer to the API project".
Comments
Comments are closed.
It's nice to see how you could do this. I don't think I would build an chatbox for the reasons as Georgiana told us. It felt as all lot of work for something that can be done with webhooks instead. As an example; last failed release/pipeline. I would expect to get this information right when it's happened and not when someone thinks, what's the status of the latest release? But that's me. ?
The talk focused on what the speaker and their team had built, how they built it and how to set it up. A lot of time talking about how to set it up, could have been replaced with 'look at the code samples and read the docs'.
There was little focus on how an AI like this could be implemented into the audiences's workflows and what benefits it would have over other forms of automation.
I would also have liked it, if the talk had started with the demos, so you'd get: "Look what we can do - here's how we did it".