Helping your bot grow a brain

Botkit Studio now come with Microsoft’s AI built-in!

Ben Brown
Howdy

--

Today, we’re launching a new integration between Botkit Studio and Microsoft LUIS. With this integration enabled, developers can seamlessly use LUIS’ natural language processing service in their bot and in the bot building tools provided by Botkit.

Try it now >

Natural language processing is an important component of the bot building stack, but using NLP is not as easy as using a database or content management system. NLP systems don’t do anything until they’ve been filled with training data and taught to recognize relevant information. They are black boxes— input come in and output comes out, but there is no way to know for certain what the results will be or why. As a result, it is hard to get started, hard to design with, and hard to know when to adopt NLP over easier — but less capable — techniques like regular expressions.

LUIS allows you to easily add conversational intelligence to your Botkit Studio project.

We like LUIS because it is the most straightforward natural language API available, and the results we’ve seen in our tests have been of a high quality. Whereas other NLP tools have expanded to add platform specific features or automatic management of replies, LUIS does nothing but understand natural language input, and it does it well.

We’ve designed this integration to ease the experience for developers who want to use NLP, so a new application can grow into a fully fledged NLP system, while also benefiting from it on day one. We have done our best to humanize the sometimes obtuse terms used in NLP — words like “utterance” have been replaced by descriptions like “example text.” Our goal is to make these powerful and exciting capabilities as accessible as possible to developers building and shipping real apps today.

Here’s a few ways we’re trying to make natural language processing easier to use for you:

Built-in Common Intents

Browsers have back buttons, windows can be closed by clicking the “X” button. Just like websites and apps, all bots need to offer some sort of basic navigational tools and user interface elements. Responding appropriately to the common greetings, complaints, and requests for help that a bot receives is critical to the user experience.

Instead of asking each of our developers to individually train their bot to recognize this type of message, Botkit Studio provides a pre-trained set of requests to LUIS that automatically provide understanding for messages as diverse as “hey you dumb robot i need some help over here,” and “who are you and why are you sending me messages all of a sudden?”

Connecting these common intents to Botkit Studio’s flexible dialog authoring tool means that bot developers can provide customized, appropriate responses for their users without thinking up 200 clever ways to say “hello.”

Train Your Brain with Real Messages

In order for an NLP tool to work, it needs to be provided with a lot of example data. Where is this data supposed to come from? A developer can only make it so far with their imagination.

Enter Botkit Studio’s message console. With the console, developers can collect, monitor and classify real messages from real users from the very first day of development. These messages can then be passed on to LUIS as training data directly from the Botkit console.

Adding training data to LUIS directly from the Botkit Studio console

A bot that starts by matching simple keyword triggers can then use the resulting messages as seed data for training LUIS. The bot’s owners can regularly review and retrain with fresh data as their users interact with the bot. The bot gets smarter and smarter the more people use it!

Use Intents from LUIS as Triggers in Botkit

In Botkit Studio’s IDE, developers can now use LUIS triggers alongside their traditional keyword or pattern based triggers. Developers can start with a simple list of keywords, and then seamlessly transition to an NLP-based solution without changing the bot’s code or configuration. By adding well-trained intents as triggers, the bot’s capacity to understand user messages gets a big upgrade.

Our integration with Microsoft LUIS is available today for bots built with Botkit Studio. Sign up for an account here, then check out this Knowledge Base article for instructions on how to get started with LUIS.

Thanks to our friends at Microsoft who made this integration possible with their generous support and assistance.

Are you building an NLP-powered Botkit bot? Let us know!

Join our 5000-member developer community to discuss your project with fellow bot builders!

--

--

Editor for

I’m a designer and technologist in Austin, Texas. I co-founded XOXCO in 2008.