ok Chatbot

Chatbot for a swiss retailer that delivers deals to customer.
Backend service for this chatbot: Apometact
Introduction blog post: Introducing the first retail chatbot in Switzerland
Lessons learned blog post: Nine key learnings after building the first Swiss retail chatbot

Features

• Understands natural language
• Displays rich elements such as cards, quick replies and images
• Implements best practice guidelines
• Translatable into different languages
• Store and retrieve content and user data from a database
• Contact dialog for human interaction

Installation development

• Copy default.json to development.json.
• Run npm install in your cli.
• Setup required services
  • mLab database
  • Facebook app
    • Set Messenger webhook to https://[subdomain].localtunnel.me/facebook/receive.
  • Facebook page
    • Connect app with page
  • Dashbot.io (optional)
  • Sendgrid (optional)
  • api.ai
    • Import api.ai ok-chatbot.zip
• Update development.json.
• Run npm run tunnel in a second cli.
• Update tunnel script command in package.json and run npm run dev in your cli.
• Go to https://www.messenger.com and start messaging with the bot.

Installation production

• Copy default.json to production.json.
• Run npm install in your cli.
• Setup required services
  • Heroku
  • Facebook App
    • Set Messenger webhook to https://[appname].herokuapp.com/facebook/receive.
    • Request messenger permissions.
  • Facebook Page
    • Connect app with page
  • Dashbot.io (optional)
  • Sendgrid (optional)
  • api.ai
    • Import api.ai ok-chatbot.zip
• Update production.json.
• Go to https://www.messenger.com and start messaging with the bot.

Design

The picture below depicts the service architecture including the backend and shows they communicate with each other.

Every message in a dialog with the bot is going through same process. This diagram shows how the services handle and process the message.