Microsoft CEO Satya Nadella talked briefly about a major update to the company’s conversational AI tools. You can now more easily create, test and improve bots that run on Azure or your own servers and work across multiple platforms.
“At this conference, we’re launching a hundred plus features for the bot framework so that you can continue to build these conversational interfaces and give them more customizations,” Nadella said.
There are now 30,000 active bots per month that use Microsoft’s conversational AI tools. They handle 30 million messages per day for a thousand companies, including Macy’s, Asiana Airlines, Stack Overflow, KPMG, Telefonica, HP and UPS.
And chances are you’ve already talked with one of those bots without realizing it because you don’t need to be using a Microsoft product to interact with a bot that leverages Microsoft’s technologies.
Microsoft’s Conversational AI tools let you deploy bots on a website, Slack, Facebook Messenger, Kik, SMS using Twilio, Telegram, Cortana, Skype for Business, Microsoft Teams, GroupMe and email.
It doesn’t work with anything from Amazon, Google or Apple though.
Microsoft’s Bot Builder SDK has been updated. It lets you pick a bot design and then create your own bot from this model. Starting today, QnAMaker is now also available as a final release. It lets you turn a good old FAQ into a set of questions and answers for your bot.
The bot emulator has been updated too to make it easier to debug your bots. There’s a new component called Dispatch that lets you dispatch queries to the right model or knowledge base. You can also manage authentication with third-party apps to interact with other apps. This is important for e-commerce websites and on-demand services.
Microsoft’s language understanding module called LUIS lets you convert speech into intents. You can now more easily integrate addresses, people and organizations in LUIS.
Finally, there are two new projects called Project Conversation Learner and Project Personality Chat. With the first project, you can feed conversations into the platform and let Microsoft use machine learning to learn new dialogue sequences. And the second project lets you create some small talk interactions to create the illusion that you’re talking with a real person.