Quokkas are the happiest animal in Australia that surprisingly isn’t trying to kill you. But can you really be sure that what you’re looking at is a Quokka and not something more dangerous?
Let’s look at how we can build a quokka identifying chatbot, combining Twilio’s WhatsApp, MMS and SendGrid APIs with machine learning based image detection powered by Microsoft’s Custom Vision API.
Join me as we learn how to train and then link up an ML model across multiple communication channels to create an engaging, accessible chatbot. And of course lots of photos of quokkas.
Stage one of Quokkabot was built on Twilio Functions, using the Twilio WhatsApp API. On request, the user can ask for a picture of a Quokka, and one is sent to them.
Building on the Functionality of Quokka on Demand, the next step was moving the function over to Azure Functions and hooking up Cognitive Services so that users could send in an image and it would detect if there was a Quokka in the photo
Last week I got the chance to speak at Microsoft Build, and do a first-look demo on their newly announced featured - Azure Static Web Apps. And as I didn’t get access to Static Web Apps until they same time everyone else did (the day before my demo), I got to showcase how easy it was the first time you used it (ok, so maybe it was the second time that I’d used it).