Going Beyond FAQ Assistants

WebinarNovember 2021

AI assistants are one of the most in-demand topics in the tech industry right now. As technology consistently improves and data becomes more available, companies strive to keep up by building their own conversational software. When built well, AI assistants provide great strategic business value and are fun to interact with. However, the majority of assistants built to date are developed using a simple set of rules or a state machine and don’t go beyond simple FAQ interactions. The result doesn’t scale in production and often provides a rather disappointing user experience.

Rasa Developer Advocate Justina Petraityte challenges the baseline approach of chatbot development by introducing machine learning-based methods for dialogue management. Justina covers the fundamentals of conversational AI, as well as machine learning behind natural language understanding and dialogue management. Finally, Justina introduces you to the Rasa Stack—an open source framework which empowers developers around the world to build their own conversational AI without any black-box tools or limitations.

You can find the slides here.

Summary

AI assistants have quickly transformed from basic notification systems to sophisticated conversational tools, largely powered by advancements in machine learning and open-source software like Rasa. Developers now have the resources to create AI assistants that go beyond simple question and answer interactions. The current AI assistants can be classified into levels, with most of them capable of answering simple questions but lacking in handling contextual conversations. Rasa aims to improve these to contextual assistants that can manage conversation flow and adapt to user preferences. Justina Petraityte, developer advocate at Rasa, stresses the benefit of in-house development of AI assistants for better customization, data security, and cost-effectiveness. The Rasa Stack, comprising of Rasa NLU and Rasa Core, offers developers tools to create scalable, adaptable AI systems. The application of machine learning in dialogue management and natural language understanding is vital, as it enables assistants to predict user intents and extract key entities from conversations. Petraityte emphasizes the value of interactive learning for enhancing AI assistants based on real user feedback, and the potential of these technologies to revolutionize user interaction with software.

Key Takeaways:

AI assistants are evolving from simple notification systems to advanced conversational tools.
Rasa provides open-source tools for developing customizable, secure, and scalable AI systems.
Machine learning is vital for dialogue management and natural language understanding.
In-house development of AI assistants allows for greater customization and data security.
Interactive learning can significantly enhance AI systems based on user feedback.

Deep Dives

Progression of AI Assistants

AI assistants have see ...
Read More

n major transformation over the past decade. Initially, they served as basic notification tools, providing reminders without engaging in true conversation. Today, they have advanced to FAQ-level interactions, capable of answering straightforward questions but still lacking the ability to manage complex dialogues. Justina Petraityte from Rasa explains that the future of AI lies in contextual assistants, which will not only answer queries but also maintain the flow of conversation and adapt to user preferences. "We believe that in the future, in a few years, we can provide the software that would enable anyone to build personalized assistants," she says, emphasizing the potential for AI to revolutionize user interaction with software.

Benefits of In-House Development

Developing AI assistants in-house offers numerous advantages, including enhanced customization, faster development cycles, and improved data security. Petraityte stresses that relying on hosted solutions like Google Dialogflow restricts customization and may not adequately address specific use cases. By building AI systems internally, companies can shape frameworks to their unique needs and retain complete control over data. "When you build your own AI, you can choose how quickly or how slow, how much time you want to dedicate for the development," she says, highlighting the flexibility that in-house development affords.

Rasa's Open-Source Framework

Rasa's open-source framework, including Rasa NLU and Rasa Core, equips developers to create AI assistants that can scale and adapt to various domains. Rasa NLU focuses on natural language understanding, enabling AI to interpret user inputs and extract relevant information. Rasa Core, on the other hand, manages dialogue, utilizing machine learning to predict the next best action in a conversation. This approach allows developers to build AI systems that go beyond pre-defined rules and adapt to real conversational data. Being open-source, Rasa Stack allows for extensive customization, making it a powerful tool for developers wanting to create sophisticated AI assistants.

Machine Learning in Dialogue Management

Machine learning plays a vital role in dialogue management, enabling AI assistants to make informed decisions about conversation flow. Rather than relying on rigid rule-based systems, Rasa uses recurrent neural networks to incorporate previous conversation states and current user inputs. This approach allows AI systems to predict the most appropriate action based on real conversational data, leading to more natural and engaging user experiences. Petraityte highlights the importance of using machine learning to close the feedback loop and continually enhance AI systems: "Interactive learning is just one of the ways how you can get the feedback and the data from real users," she explains, emphasizing the potential of machine learning to enhance AI capabilities.