Ruby framework for conversational AI

Create intelligent, conversational AI experiences with a framework designed to maximize developer happiness.

Integrates with

Facebook Messenger
Microsoft LUIS
Google WaveNet Voice

Conversational AI Stack

Xip Kit is a "batteries included" framework. With it, you can utilize the entire ecosystem of AI services and extend it to create powerful, robust AI experiences.

First-Class NLP
NLP intent and entity detection is a core part of controller replies. Just pick an NLP service, train your model, and you're done.
Plug and Play Components
The AI ecosystem changes rapidly. Xip Kit offers a variety of pre-built components or you can build your own.
Developer Happiness
From the Ruby programming language, to hot-code reloading, Dev Jumps, and built-in tunnels, Xip Kit helps you launch quickly.
Ownership and Control
The AI experience you create is yours to own and extend. Your investment is your competitive advantange, not ours.
MVC Architecture
The easy-to-learn architecture is familiar and ensures your entire team can stay focused on building the AI experience.
Utilize RSpec throughout for testing. You can even use your existing continuous integration pipeline.
Cross-platform Replies
Replies work seamlessly across all messaging platforms. Platform-specific replies can also be sent automatically.
Data Models
You can use an ORM, REST API, GraphQL, or anything else to load and store your data. It's all just Ruby!

Smart Assist

Smart Assist elevates the user experience beyond each individual AI component. Each feature is designed to enable your AI experience to adeptly handle real-world scenarios.

Catch-alls automatically trigger when a user says something unexpected or the bot experiences an error. They are multi-level and so they provide a graceful way of handling errors throughout your bot.
Interrupt Detection
Some messaging platforms don't support typing indicators and so it's common for users to type while a reply is being generated. Interrupt detection automatically handles these cases.
Unrecognized Messages
When a user says something unexpected within their context, this special controller fires before catch-alls. This provides a powerful way for you to handle NLP globally for your bot.
Homophone Detection
Some English letters sound like other words. So for example, pronouncing "C" sounds like "see" or "sea". When users message your bot from their smart assistant, these details matter.
Platform Errors
Our message platform components automatically translate platform errors into Ruby callbacks. So when your bot mistakenly attempts to text a landline, you can take the appropriate action.
Conversational UX
Great converstional UX is a priority for us. Automatic delays add natural, human pacing to your replies. We've even managed to add quick replies in SMS/Whatsapp messages to reduce chat fatigue on those platforms.

Xip Kit is made possible thanks to generous contributions from our sponsors and backers. Learn more about how you can become a sponsor.