How to Build a Quokkabot
summary
A WhatsApp and email chatbot that identifies quokkas by sending images to Microsoft Custom Vision and returning a confidence score builds on Twilio WhatsApp, Custom Vision, SendGrid and Azure Functions. Training starts with ~50+ images per tag; early errors confuse wombats and kangaroos but accuracy reaches ~99% on familiar photos (dog classified 98% as not-quokka). Replies include a quokka photo or fact, results stream into Twilio Sync for a live dashboard, and the demo runs live with occasional retraining needed.
AI-written summary · about
chapters (9)
- 0:00Start
- 14:57Image Classification
- 17:33Custom Vision
- 19:00Classify Image
- 30:46Quokka Facts
- 34:40Fun Facts about Quokkas
- 43:03Email
- 49:25Update Image Function
- 53:37Add a Twitter Bot
related talks
57mHow to Build a Quokkabot
20mHow to Build a Quokkabot
01h 01mBuilding fun and creative messaging experiences on WhatsApp
54mChatbots: The hype, challenges and opportunities
48mAn App That Sees, Hears and Responds
29mCandy Crush QA AI Saga
35mAI JavaScript Rocks
59mHow to create No Code Chatbot with Azure QnAMaker
