Redis and Asurion: How to build for fast customer support using RAG (DAT203)

Introduction to Redis

• Redis is a popular database known for its caching capabilities, trusted by over 10,000 customers in production.
• Redis has a large developer community and is used across various industries like media, financial services, e-commerce, and more.
• Redis has evolved to meet the needs of customers beyond caching, offering features like novel data structures, a query engine, event streaming, and time series capabilities.
• Redis provides tools and integrations to help developers work with the database, such as Redis Insight and Redis Co-pilot.

Aswan's Journey with Redis

About Aswan

• Aswan is a large tech company with 300 million customers worldwide and over 20,000 employees.
• Aswan's main business is engineering consumer electronics and providing customer support.
• Aswan has a subsidiary called Simpler, which focuses on building a flexible customer support platform using AI.

Challenges Faced

• Aswan experienced rapid growth, leading to spikes in customer support volume that were difficult to handle.
• The existing solution of storing data in-memory was becoming harder to maintain as the scale increased.
• Aswan needed a solution that could scale, provide speed, and offer flexibility to handle their growing customer support demands.

Adopting Redis

• Aswan chose Redis due to its seamless integration with their AWS-hosted services, the flexibility of the Redis Vector DB, and the speed and latency benefits it provided.
• Redis Vector DB allowed Aswan to design flexible schemas and indexes to store metadata and provide more relevant responses to customers.
• Aswan leveraged Redis' semantic caching capabilities to reduce response times, improve customer engagement, and save costs by bypassing their entire framework.
• Redis also enabled Aswan to effectively manage chat session state and context, which was crucial for their conversational AI-powered customer support system.

Impact and Learnings

• Implementing Redis-based semantic caching resulted in a 50% reduction in response times and a significant increase in customer engagement and stickiness.
• Aswan learned the importance of starting small, validating the customer experience, and then scaling their solution using Redis' flexibility and scalability.
• Key learnings include the need for flexibility in data storage and retrieval, the benefits of in-memory semantic similarity, and the ability to scale without being constrained by the underlying technology.

Future Plans and Advice

• Aswan plans to continue simplifying their architecture, further reducing latency, and optimizing costs while maintaining a focus on delivering an exceptional customer experience.
• The team emphasized the importance of having a diverse toolset beyond just a vector database, including message brokers, caching, externalized memory management, and more advanced routing capabilities.
• Redis announced the upcoming availability of Redis Flex, which can cache and store 5x or more data at the same price as traditional RAM-based solutions, potentially addressing the growing demand for large-scale caching and storage.