We're excited to announce a new addition to our collection of open source AI assistants: the Retail starter pack. The Retail starter pack is an example chatbot for ecommerce customer service. It can assist with tasks like checking inventory on a product or the status of an order. Like the Rasa starter packs for Financial Services and IT Helpdesk, the Retail starter pack is free and available for the community to download, modify, and extend.
The Retail starter pack comes pre-loaded with training data and several retail customer service skills. Developers building assistants for retail customer service can use the code as a reference or fork the project to build directly on top.
In this post, we'll take a tour of the Retail starter pack's features and show you how you can get started building your own assistant for retail.
What the Retail starter pack can do
The Retail starter pack includes a set of out-of-the-box features common to retail customer service operations. These include:
- Looking up the status of an order via the customer's email address
- Cancelling an order
- Initiating a return
- Checking whether a product is in stock in a particular size or color
- Subscribing the customer to updates
The Retail starter pack recognizes three email accounts when retrieving order statuses or initiating a return, which are checked against the assistant's database:
The Retail starter pack serves as a working example for developers implementing similar features in their own projects. For instance, the assistant connects to an SQLite database containing product inventory and order data. It retrieves and updates records in the database when the customer requests a status update or changes the order. The assistant also includes multiple forms which are used to guide the user through conversation flows, as well as retrieval intents for handling FAQs and chitchat. Lastly, Duckling is used to extract email address and number entities.
Getting started with the Retail starter pack
You can find the Retail starter pack on GitHub, where it's available to clone or fork. In the project's README, we've included detailed instructions for running the assistant, which is based on Rasa Open Source 2.1.2.
Once you've downloaded the project locally, you can continue to develop it further. Customize the assistant to fit your requirements, and continue to add skills and training data to make the assistant more robust. We've kept the Retail starter pack simple to make it straightforward to understand and customize, but developers building retail customer service chatbots can expect to cut their development time considerably by using the Retail starter pack instead of coding from scratch.
The Retail starter pack is our latest effort to provide the community with feature-rich open source examples for some of the most common use cases developers build with Rasa. If you're looking for more Rasa examples, check out our other starter packs for IT Helpdesk support and Financial Services. Lastly, we're always looking for new ideas for future starter packs. If there's a use case you'd like to see, let us know!