Meetup notes: Microsoft Bot Framework

Last night, I went to a meetup at Slack on the Microsoft Bot Framework. It was a good opportunity to get a grasp on what the Bot Framework is, and how developers are using it to create bots that deploy on several different channels, including (as of March 2016):

  • Text/SMS
  • Office 365 mail
  • Skype
  • Slack
  • GroupMe
  • Telegram
  • Facebook Messenger
  • Kik
  • Web (via the Bot Framework embeddable web chat control).

The general idea is that you can write a bot in the Microsoft Bot Framework, and it will be portable to all these channels, with some slight modifications depending on the specifics. Great! But what is a bot, anyway?

Generally, a bot embodies some kind of AI, or artificial intelligence. It’s an app that acts like a person, and it helps you accomplish work by using contextual meaning. If you’re an iPhone user, you’re familiar with Siri. Siri is essentially a bot – it asks you questions and accomplishes some kind of work by reasoning about your answers. Here’s a screenshot from the Bot Framework splash page:

Screen Shot 2016-06-23 at 1.28.01 PM

Mat Velloso gave the presentation, and one example he gave is called Murphy. You ask Murphy any kind of ‘what if’ question, and it goes out and finds images that match your scenario, then mashes them together. Basically, it’s face swap via bot. You can say ‘what if I were the president?’ and Murphy will probably give you an image of President Obama with your face. This is a rudimentary example, but you can see how bots would make lots of things faster, especially on a mobile device. You can give short, partial answers and depend on the natural language processor to figure out what you mean.

The SDK, available on GitHub, is written in C# and Node.js. You can fork the repo and start working from examples to build your own bots. Here are a few more resources that were covered during the presentation:

docs.botframework.com – take a look at the docs

luis.ai – lets your app understand language

microsoft.com/cognitive-services – a whole suite of services like LUIS, including face and voice recognition, academic knowledge, image search, etc.

bot framework emulator – lets you play with your bot as your develop it. Includes a GUI for Windows and a command-line version for Mac / Linxu.

 

Looking forward to playing with this framework and maybe building a bot of my own in the near future!

Downgrading a DigitalOcean Droplet with ServerPilot & WordPress

If you are a WordPress admin looking for a way to downgrade your DigitalOcean droplet, you probably came across their article How To Downgrade DigitalOcean Droplets, which tells you to use rsync to move files from your old droplet to the new droplet. If you are using using ServerPilot to manage your DigitalOcean server, do not follow these instructions. ServerPilot has an article called Cloning and Resizing Servers that explicitly tells you not to use rsync. There are a few problems with rsync – basically, it will not correctly clone your ServerPilot configuration, so your new site will not work. Here’s a more automated method using a WordPress plugin called ManageWP, which I found to be a better solution:

  1. Create a new server on Digital Ocean – do this using the smaller subscription that you want to downgrade to.
  2. Connect the new server to Server Pilot
  3. Install WordPress with one click on the new server
  4. Install Manage WP on both sites (you need to install the plugin under the same account in both sites – just do the sections titled Install the ManageWP Worker Plugin and Set Up ManageWP).
  5. Migrate WordPress with Manage WP 
  6. Go to the new IP address in a browser and verify that it works
  7. Update DNS records with GoDaddy to point to the new site

If you found this useful or have any feedback on this process, leave a comment!

GitAchieve

GitAchieve allows you to compete with other GitHub users to improve productivity through gamification.

GA-demo-1-1

 

GitAchieve was developed by a team. As Product Owner, I led the team in identifying key features and managing the backlog, as well as contributing on several key engineering tasks:

● Designed RESTFul API in Node/Express to serve custom GitHub datasets to the frontend
● Formulated Postgres SQL queries to persist relational user data
● Set up continuous integration with TravisCI for code quality and rapid iteration
● Leveraged Gulp automation to build, test and deploy production code

GitAchieve is deployed to an EC2 instance on AWS.

Check out the GitHub repo