Rasa’s Community Showcase: What Developers are Building with Rasa

Developers use Rasa to build an incredible variety of AI assistants—everything from voice-enabled search, to customer service chatbots, to bot-builder platforms that use Rasa as their core technology. In the community, we also see fantastic, creative conversational AI projects that developers have built just for fun.

We launched our new Community Showcase to highlight the range of projects built with Rasa and to help developers get inspired by what’s possible. We’ve included 20+ AI assistants in this first version of our showcase, representing both commercial and personal projects.

Think of the showcase as a gallery of assistant profiles, demonstrating all the capabilities of each assistant, at a glance. A picture might be worth a thousand words, but we think video is even better, so each assistant’s profile includes a video demonstration. We also include relevant details like messaging channel, languages supported, and if applicable, a link to the assistant’s open source code. If the assistant is publicly accessible, we include a link for you to try it out yourself.

We want the Community Showcase to be the most complete collection of AI assistants built with Rasa, all in one place. Our goal is to grow the showcase over time, bringing in more and more examples. To do that, we need our community’s help! If you’ve built an assistant with Rasa and it’s running in production, whether for work or for a personal project, we’d love to include it.

In this blog post, we’ll introduce you to a few of the chatbots and voice assistants you can find in the Community Showcase. Then, we’ll outline how to get your assistant featured.

Meet the Assistants

The AI assistants in the Community Showcase span an amazing variety of use cases and industries, but they have a few things in common. They’re all running in production today, and they all use Rasa. Get to know a few of them here:

Hey Sensei

Adobe’s Hey Sensei stands out for its voice-enabled interface and intuitive approach to filtering Adobe Stock image assets. Users can find just the right image among the hundreds of millions of photos on Adobe Stock, simply by asking for what they want.

Hey Sensei understands commands like “Find photos of a couple on vacation,” but the real power of the assistant is its ability to build on previous requests to further refine the query. Let’s take the vacation photo search as an example: you could then add qualifiers like “Add a beach,” “Add a sunset,” and “More blue tones.” The end result? A selection of stock photos depicting just what we want: a couple on a beach vacation at sunset, with lots of blue tones in the image.

Hey Sensei by Adobe

Check out Hey Sensei in the Community Showcase, and also be sure to watch Brett Butterfield, Director of Engineering at Adobe Sensei, present his talk Powering Search Using Natural Language, Machine Learning, and AI from the Rasa Developer Summit.

Dialogue Virtual Clinic

Canadian telemedicine provider Dialogue developed the Dialogue Virtual Clinic AI assistant to automate the patient intake and evaluation process, allowing human care providers to spend more quality time with patients. The assistant gathers information about the patient’s chief complaint through a series of questions before transferring the patient to a human health care provider. As a result, health care providers are able to begin the virtual examination with all the relevant background information they need, saving time and allowing providers to serve a greater number of patients.

The Dialogue Virtual Clinic includes a number of interesting technical features that extend the capabilities of Rasa. For example, the assistant supports rich media like image uploads, and patients can click a visual representation of the human body to indicate where they’re experiencing discomfort. Dialogue also customized the Rasa Form Policy to integrate an external inference engine for managing the patient evaluation dialogue.

Virtual Clinic by Dialogue

See the Dialogue Virtual Clinic in the Community Showcase, and also read the case study for more details.

Djingo

Djingo (created by Orange, a European telecom provider) is a great example of an AI assistant providing advanced, automated customer service. Orange built Djingo to handle first line technical support for phone and internet, walking customers through troubleshooting steps to resolve connectivity issues.

The chatbot supports multiple messaging channels, including Skype, Facebook Messenger, and webchat. Since deploying Djingo, Orange has deflected support contacts from human agents while increasing customer support availability after hours.

Djingo by Orange

Visit Djingo on the Community Showcase and learn more in the case study.

Anna, a Virtual Assistant for Kids

Our next AI assistant was built by Julian Gerhard, a hero contributor in the Rasa community who built Anna in his free time.

Anna is a proof of concept that illustrates how to use a SmartBoard in an educational setting. The assistant is a voice-controlled avatar that uses Rasa for NLU and dialogue management, running on a Raspberry Pi connected to a touch screen. The avatar responds to voice commands to move different parts of its body and also plays a game called “What means.” The user can ask the avatar what any word means, and the avatar will query Wikipedia to explain the definition.

Anna by Julian Gerhard

Watch a demonstration on the Community Showcase.

Neon

Lastly, we have Neon, created by the bank N26. Neon handles 20% of all support requests for N26 through the bank’s mobile app. The chatbot is capable of performing customer support tasks as simple as updating contact details, and as complex as reporting a lost or stolen card.

That would be impressive enough in one language, but Neon does it in several, including English, French, and German. To manage this complexity and allow developers and content creators to work together more effectively, N26 integrated Contentful and created a wrapper for writing custom actions. This allows team members who aren’t developers to manage content and add actions without writing code.

Neon by N26

Check out Neon on the Community Showcase and go deeper with N26 data scientist Edouard Malet’s presentation from the Rasa Developer Summit: Building Scalable Chatbots by Empowering Content Creators.

Get Featured!

We hope you’re feeling inspired by the innovative examples we’ve mentioned here, and if you’ve built an AI assistant with Rasa, we’d love to add yours to the group. We’re looking for AI assistants that:

  • Use Rasa for NLU and/or dialogue management
  • Are running in production. If your assistant is currently under development, check back once you’ve launched!
  • Include a video demonstration. Be sure to film a screen capture or video of your assistant engaging in conversation.

Submit Your Assistant

To add your assistant to the showcase, send us an email at community-showcase@rasa.com. We’ll gather a few details about your assistant, collect a video recording, and work with you to create a showcase page profiling your project. If you’ve given a talk about your assistant or written a tutorial, those are great resources to share along with your profile, but they’re not required to be featured.

Aside from those guidelines, we welcome all kinds of use cases and conversational AI projects, no matter how big or small. We love seeing the assistants developers have built just for fun, to learn a new skill, or to demonstrate a proof of concept.

Have more questions about the criteria? Here are a few answers to commonly asked questions:

Q. My chatbot isn’t in English. Is that okay?
A. Yes! Non-English submissions are both welcome and encouraged.

Q. My assistant is gated behind a login, or for internal use at my company. Can I still be featured?
A. Yes! But we do need a video demonstrating your assistant’s capabilities. If your assistant is not gated (i.e. the assistant is accessible over the public internet), it’s an added bonus to include the link so we can direct Community Showcase visitors to try it out.

Q. My assistant isn’t a traditional text-based chatbot. Does that count?
A. Yes! We’re looking for all types of conversational AI projects, including voice-based assistants and platforms that are powered by Rasa. The more outside the box, the better.

Don’t see your question here? Send us an email at community-showcase@rasa.com.

Conclusion

We’re excited to continue expanding our Community Showcase, highlighting the conversational AI projects developers are building with Rasa. With 20+ AI assistants and counting, we’re on our way to building a central place for developers to see the full spectrum of Rasa chatbots and voice assistants.

Visit the showcase to get inspired by more AI project ideas. Then, take the next step and show off what you’ve built by submitting your assistant!