Looking for a productive way to spend some holiday downtime? We've crafted a flexible learning journey to help you level up your conversational AI skills with Rasa Pro. Whether you're new to Rasa or looking to expand your capabilities, this curated path will take you from foundational concepts to advanced features. Take it at your own pace – it is the holidays after all!

Before You Begin

To make the most of this learning journey, you'll need:

Required:

A Rasa Pro developer license
Basic Python programming knowledge
Git (for accessing example repositories)
Command line familiarity

Nice to Have:

Experience with machine learning concepts
Familiarity with conversational AI basics

Day 1: Begin with the Basics - Building Conversational AI with CALM

Preliminary time estimate: 2-3 hours

Your journey starts with CALM (Conversational AI with Language Models), Rasa's approach to building AI assistants combining the power of large language models with precise business logic. CALM allows you to create more intelligent and responsive AI assistants while maintaining control over their behavior and costs.

To start building your conversational AI assistant with Rasa, dive into the Building Conversational AI Assistants with CALM video series. This series covers all the foundational concepts you need to get up and running with Rasa Pro, guiding you through using CALM to create more intelligent and responsive AI assistants.

As you work through the video series, you'll also find the accompanying CALM documentation helpful for deepening your understanding. This guide will explain how to set up and integrate CALM into your Rasa projects, offering valuable insights into the specific tools and features you'll need to build your AI assistant.

Follow the Rasa Pro text tutorial to fully apply what you have learned. This tutorial provides step-by-step instructions for building your own AI assistant from scratch, reinforcing the concepts learned in the video series and documentation while offering practical hands-on experience with Rasa Pro.

By combining the video series, CALM documentation, and tutorial, you’ll understand Rasa Pro's theory and practical implementation, setting a strong foundation for developing conversational AI assistants.

Watch the Video Series and Follow Along with the Text Tutorial: Building Conversational AI Assistants with CALM - YouTube Playlist
Explore the Documentation: CALM Documentation
Follow the Step-by-Step Tutorial: Rasa Pro Text Tutorial

Day 2: Enhancing Your AI Assistant - Cutting Costs with Business Logic

Preliminary time estimate: 3 hours

Now that you understand the basics let's optimize your assistant. Today, we'll explore how integrating business logic with large language models (LLMs) can significantly cut costs and improve your AI assistant's efficiency:

Read the Blog Post:Cutting AI Assistant Costs: Enhancing LLMs with Business Logic

This blog post discusses how separating conversational abilities from business logic execution, as done with the CALM approach, can significantly reduce costs and improve response times (compared to fully unstructured or semi-structured approaches like LangChain/LangGraph). The article compares these approaches' performance, latency, and cost, highlighting how CALM achieves up to 77.8% cost reduction and better reliability, even when handling complex user inputs.
Watch the Live Stream:Enhancing LLMs with Business Logic - Live Stream

The live stream complements the blog post by diving into practical examples and offering further insights into how you can implement business logic with LLMs for a more efficient and scalable conversational AI system.
Explore the GitHub Repository:CALM LangGraph Customer Service Comparison - GitHub Repository

The GitHub repository provides the code used in the blog's experiments, showing how CALM and LangGraph perform in a customer service context. You can explore and try the code for yourself to better understand the differences in performance and cost-efficiency between these two approaches.

Day 3: Take a Breather - Understanding Evaluation

Preliminary time estimate: 1 hour

Today, we're keeping it lighter—it is the holidays after all! While building assistants is exciting, it's equally important to know how to evaluate their performance. This shorter reading will help you assess whether your chatbot is actually doing what it should.

Read the Blog Post:Recipe for Comparing Chatbot Implementations

This blog post outlines a practical approach to evaluating chatbot implementations. It covers the importance of establishing success criteria for specific tasks your assistant is designed to perform (e.g., booking appointments, answering queries). The blog emphasizes error analysis, performance benchmarking, and user testing as critical methods for assessing chatbot effectiveness. It encourages a structured approach to comparing different implementations by setting clear goals and measuring success against these goals.

Day 4: Building Reliable Agentic Bots with LLaMA 8B

Preliminary time estimate: 3-4 hours

Ready to level up? Today, we're diving deeper into building more reliable agentic bots using powerful models like LLaMA 8B. This day focuses on creating bots that handle conversational tasks and exhibit a higher level of reliability and agentic behavior.

Read the Blog Post:Reliable Agentic Bots with LLaMA 8B

In this blog post, Rasa explores how to build more reliable and agentic bots using LLaMA 8B, a powerful language model. It highlights the steps for integrating this model into your assistant to ensure the bot can handle complex interactions with greater reliability and decision-making capability. The post covers key insights into how the LLaMA 8B model can be optimized for robust conversational flows, making it an ideal choice for creating dependable, task-driven bots.
Watch the Live Stream:Reliable Agentic Bots with LLaMA 8B - Live Stream

The accompanying live stream provides a deeper look at the blog's concepts, featuring practical examples and demonstrations of how to build agentic bots using LLaMA 8B. It's a great resource for seeing the model in action and understanding how to implement it for high-performance chatbot solutions.

Bonus Challenge: Voice Assistance and IVR

Preliminary time estimate: 2 hours

Ready for an advanced challenge? Let's explore the world of Voice Assistants and IVR (Interactive Voice Response). These systems allow users to interact with bots through voice, making them more accessible and engaging. This bonus section looks at how you can integrate natural language processing (NLP) into voice applications.

Read the Blog Post:Natural Language IVR

This blog post explains integrating natural language understanding (NLU) with Interactive Voice Response (IVR) systems. It covers how Rasa can replace traditional rigid, menu-based IVR systems with more flexible, natural language-driven solutions. The post discusses the benefits of using conversational AI for IVR, such as improving the user experience and enabling more intuitive interactions.
Explore the Documentation:How to Start Building a Voice Assistant

The Rasa Pro documentation provides a detailed guide to building a voice assistant. It covers the necessary steps for setting up a voice interface with Rasa and integrating various voice technologies and frameworks. This resource is ideal for learning how to add voice capabilities to your assistant, whether for IVR applications or broader voice interactions.

Coming Soon: Voice Assistant Tutorial

Keep an eye out for our upcoming voice assistant tutorial that will guide you through setting up your voice-enabled assistant. This live tutorial will walk you through practical steps and examples, ensuring you understand how to integrate voice features seamlessly into your assistant. We'll provide details and a link once it's available!

Congratulations on exploring this learning journey! Remember, you can always revisit these resources and take them at your own pace. The holiday season is about enjoying yourself while learning something new. Have questions or want to share your progress? Join our community to connect with other developers on their Rasa journey.

Happy learning and happy holidays from the Rasa team! 🎄✨