What is the architecture of a Virtual Assistant?

For Virtual Assistant Architecture, please see Virtual Assistant Architecture.

Azure Bot Service

Related questions and answers

  • Yes, aspects of each Virtual Assistant skill (Language Model, Dialogs, Integration Code and Language Generation) can be customised by customers and/or partners. The full source code is provided on GitHub.

  • The Bot Framework SDK provides an open architecture. You can build, connect, deploy and manage intelligent bots to naturally interact with your users on a website, app, Cortana, Microsoft Teams, Skype, Slack, Facebook Messenger and elsewhere.

  • Botskills command line tool allows you to automate the connection between the Virtual Assistant and your Skills, which includes the process of updating your dispatch models and create authentication connections where needed.

    Learn more: https://microsoft.github.io/botframework-solutions/reference/skills/botskills/

  • Azure Bot Service allows you to create all kinds of bots, from internal corporate bots that help complete specific transactions to public-facing bots in channels like Skype, Facebook Messenger, Slack and Microsoft Teams that represent your brand and help improve customer service. Bots can be infused with the breadth of Azure cognitive services to enable vision, speech, text analytics, sentiment analysis and others. When necessary, bots can pass the conversations to human agents.

  • To set up Enterprise Notifications for a Virtual Assistant, please check out: https://microsoft.github.io/botframework-solutions/howto/samples/enterprisenotifications/

  • QnA Maker extracts and maintains question-and-answer pairs. Language Understanding (LUIS) identifies intents and extracts entities. With Bot Framework SDK V4, additional tools help you route utterances between them, so you can easily combine both in a bot.