How to create a great AI Assistant

Introduction

Conversational experiences are critical in today's era of omnichannel customer support. However, it's not always an enjoyable experience; we've all been on hold way too many times only to find out that the information we needed was available to us the whole time through an easier path. Today, I am happy to invite John, CEO of Ryfi, who is helping companies create great conversational AI assistants. John, it's so great to have you here with us today. Could you tell us a little bit about yourself and about Ryfi?

Introduction to Ryfi

Absolutely, my pleasure to be here as well. I’m John Matthew, the co-founder and CEO of Ryfi Technologies from India. Think of us as a consumer brand—we have more than 40 million users in 23 global languages. We’ve been partners on the AI front for the last 10 years, starting our AI journey back in college in 2011. Over the years, we have perfected the technology and created an AI platform called Essie, which drives conversational AI, enterprise search, and more.

Attributes of a Great AI Assistant

A great AI assistant, from a consumer perspective, should be easy to use, smart, and available on the user’s preferred platform—whether that’s messaging, email, or phone calls. On the enterprise side, it should be easy to build, operate, and scale, while being capable of generating revenue and reducing costs.

To meet these goals, it’s crucial to address three key perspectives:

Training efficiency and speed
Converting content discovery into actual commerce
Providing a consistent experience across all platforms by integrating structured and unstructured data.

The typical problems with rule-based or intent-driven chatbots can be overcome by creating a smart AI system that evolves with data changes and can proactively reach out to customers to complete transactions.

Case Study: Federal Bank

The journey with Federal Bank began in 2019 after they tried around 30 different chatbot providers. The main issues areas included data training and keeping the system updated with regulatory changes. With our platform, we started with a simple website-based chatbot and expanded to multiple platforms like WhatsApp, Messenger, Google Search, and voice assistants. Over time, we have been able to achieve a 98% accuracy rate and have even enabled core banking functions via the AI assistant.

Core Banking Operations

We initially determined 18 core banking operations based on user needs and the feasibility of integrating existing APIs. The implementation process involves handling conversational sequencing and integrating Google’s Dialogflow with our proprietary text and data magic systems to manage user context and trigger API actions.

Maintaining Context

Keeping track of context across platforms and allowing users to switch contexts seamlessly is vital. Our system maintains the entire conversation sequence, even across different platforms, ensuring that context is always preserved and interactions remain smooth.

Future Enhancements

The future for Ryfi revolves around on-device processing for hyper-personalization while maintaining large-scale data processing on the cloud. This approach ensures user data privacy and security while offering a personalized experience.

Where to Learn More

Federal Bank, the fourth largest private sector bank in India, provides an online assistant deployed on their website. Google also conducted a case study on this implementation. Additionally, there’s a podcast featuring insights from Federal Bank’s VP of Digital, Mr. Jidesh, alongside John.

Conclusion

Thank you, John, for providing us with a deep dive into Ryfi’s AI assistant, Essie, and its architecture. We have gained valuable insights into developing conversational AI assistants that deliver consistent and efficient customer support across multiple platforms.

Keywords

Conversational experiences
Omnichannel support
AI assistants
Ryfi Technologies
Federal Bank
Context preservation
Hyper-personalization
On-device processing
API integration

FAQ

Q: What makes a great AI assistant? A: A great AI assistant should be easy to use, smart, and available on the user’s preferred platform. It should also be easy to build, operate, and scale for enterprises, while being capable of generating revenue and reducing costs.

Q: How did Ryfi help Federal Bank with their AI assistant? A: Ryfi started with a simple website-based chatbot and expanded it to multiple platforms. They achieved this by integrating Google’s Dialogflow and their proprietary technology, which allowed Federal Bank to process core banking functions through the AI assistant.

Q: What are the future plans for Ryfi’s AI technology? A: Ryfi aims to miniaturize their technology for on-device processing, ensuring hyper-personalization while maintaining large-scale data processing on the cloud for enhanced privacy and security.

Q: How does Ryfi handle context switching in conversations? A: The system maintains the entire sequence of conversations across different platforms, ensuring that the context is preserved and allowing for seamless interaction even when the user switches topics or platforms.

Q: Where can I learn more about Federal Bank’s AI assistant? A: You can visit Federal Bank’s online assistant on their website and check out the case study done by Google, or listen to the podcast featuring insights from Federal Bank’s VP of Digital and John from Ryfi.