At KITT.AI we have a vision that "Conversational Understanding" should be an infrastructure service available to every developer.

Five months ago, we launched our first product, a customizable wake word detection engine, which is now being used by more than 3000 developers. Then what's next?

Right before Thanksgiving 2015, we began our quest to creating a beautiful and elegant framework that helps developers design and program natural language dialogue systems easily, and add them to their devices, apps and products quickly. We have created what we think is the best conversational engine that we want to put into the hands of every developer. Today we're very excited to announce the private beta of ChatFlow - a multi-turn, multi-bot, multi- platform framework for creating conversational agents.

ChatFlow incorporates three elements that are crucial to make AI-embedded conversations delightful for customers:

1. Multi-turn dialogue

How many times have you seen demos like "book me a four-star hotel in Seattle under $300 with free WiFi less than 2 miles from the city center for tomorrow night"? Conversational AI wouldn’t be as big a deal if we all spoke like that.

In the real world, people book hotels with agents by engaging in a dialogue, not a long one-way monologue. It takes multiple turns, with validations and confirmations along the way.

With ChatFlow, we wanted our framework to handle these multi-turn dialogues with ease, distilling years of research from the SigDial community. We built a dialogue manager into ChatFlow that handles dialogue context, tracks dialogue state and executes dialogue rules.

2. Natural Language Understanding

In 2014, wit.ai (Facebook) democratized Natural Language Understanding (NLU) and brought intent classification and entity extraction to the masses. Many similar services have followed the path of wit.ai, such as luis.ai (Microsoft), api.ai (Google) and nlu.mix (Nuance). Today, NLU is a commodity.

We wanted our framework to handle all of the different NLU services in the same way. No matter what kind of .ai NLU a developer uses, it can quickly be incorporated into a multi-turn dialogue system using ChatFlow. We provide our own deep-learning-powered NLU service that can easily generalize even from limited inputs.

3. Usability and Multi Platform

Traditionally, dialogue design starts with flows on a whiteboard. However, this approach becomes increasingly painful as the dialogue becomes bigger and it's extremely hard to synchronize drawn diagrams with actual code. Running out of space on the whiteboard is the least of the problem!

ChatFlow liberates developers to focus on the dialogue flow, rather than the code. It uses a simple drag-and-drop interface that visually describes a dialogue AND at the same time implements the flow that can be executed on the server as the dialogue is designed. Design and implementation are synchronized: ChatFlow provides the full stack for programming modern dialogue systems.

Currently ChatFlow supports the following conversational platforms:

  • Amazon Alexa
  • Facebook Messenger
  • Kik
  • Layer (work in progress)
  • Skype
  • Slack
  • Telegram
  • Twilio

Great software comes with excellent documentation. We prepared more than a dozen tutorials to help you get started. Also, each tutorial comes with a video walk-through!

AVAILABILITY

You can sign up for the ChatFlow's beta program at https://chatflow.kitt.ai. Full documentation is available at http://docs.kitt.ai/chatflow/.

We are adding people to the beta in batches over time to make sure we can provide a robust service backend for your game changing conversational applications.

What's the next chapter for KITT.AI's quest in providing Conversational Understanding as a Service? We are working on something unique and innovative. Stay tuned!