The Art of Coding a Conversation: Designing Bots
summary
Argues that bots are just apps and shows concrete tooling and metrics: uses LUIS for intent/entities and the Bot Framework connector to host a single endpoint across Messenger, Skype and others; cites XiaoIce-like bot with ~40 million users, 23-exchange average conversations (about 10× industry average), Japan reach ~20% of population, and 170k US users with ~7k hour-long chats. Demonstrates Adaptive Cards that render rich, cross-platform carousels, calendars and validation as JSON templates to reduce typing and guide users.
AI-written summary · about
chapters (14)
- 0:00Start
- 2:43Command Line
- 3:15Menus
- 6:06Ubiquity of Communication Channels
- 23:38Open Graph
- 23:53Benefits of templates
- 24:39Challenges with templates
- 33:57Adaptive Cards
- 37:05Help Desk
- 40:02Microsoft Bot Framework
- 43:36Menu breakdown
- 44:44Canvas/input mismatch
- 47:01New Docs
- 47:31Parting Words of Wisdom
related talks
55mConversations for the masses
41mBuild a great conversationalist
55mEnterprise Chatbots (using messaging and AI to build novel user interfaces for work)
54mChatbots: The hype, challenges and opportunities
35mDesigning Experiences for Chatbots
01h 06mHow to Build Conversational Apps and Multichannel Intelligent Bots
42mLet’s Chat-Bots using Azure AI
54mBuilding advanced ChatBots