How a Leading Financial Data Company Commercialized AI in Under a Year

Summary

A leading US financial data analytics company needed to enable their engineering teams to build and scale secure, reliable AI applications, fast. Weaviate’s AI-native, open source vector database came with robust out-of-the-box features, flexible deployment options, strong performance, and reliable support, making it the company’s vector database of choice. In less than one year, this company commercialized AI and empowered every internal employee to ask questions about their files in their proprietary chat tool.

Challenge

A Lead System Engineer (LSE) was tasked to determine which vector database would best enable engineering teams across the company to build and scale secure applications in production. As AI excitement spread across the financial data industry, he noticed that within a month there were four to five different vector databases in use across internal teams. He needed to find and implement the right solution before tool sprawl got out of hand and his team would be stuck supporting a bloated tech stack. Additionally, competitors were starting to build their own AI tools to support faster and more accurate decision making – the company needed to move quickly to maintain the best-in-class service they provide their customers.

Why Weaviate

Early in his research, the LSE learned that his company’s engineers weren’t just looking for a database to support basic search features – they wanted to build chatbots, integrate multiple data sources and large language models (LLMs), have support for vector search, metadata, hybrid search, reranking, and filtering. He worked across teams to determine the importance of ease of implementation, performance benchmarks and infrastructure requirements like backups, restoration, and being able to deploy a vector database on premises or in their own AWS cloud environment. Many vector databases could meet multiple requirements, but only Weaviate could meet them all.

“Experimenting is one thing, but when you’re building long-term, client-facing enterprise applications, you want the right vector database with the right level of support. Weaviate was that database for us,”

Lead System Engineer from a top US financial data analytics company

Solution

The company chose Weaviate because it fulfilled four primary requirements:

• Robust feature set: Weaviate’s modular architecture allowed easy integration of various data types and sources and popular LLMs. Out-of-the-box features like hybrid search, reranking, and filtering allowed developers of various skill levels across teams to focus on building applications instead of creating those functions from scratch.
• Flexible deployment options: The company could deploy Weaviate on AWS EKS through a pre-vetted Kubernetes blueprint for internal usage, fulfilling security requirements and allowing teams to move quickly. As an open-source vector database, Weaviate could be deployed on premises and in the cloud allowing strict data control.
• Meeting required benchmarks: The infrastructure team surveyed internal teams about the technical requirements for various use cases – including queries per second, speed of implementation, and high availability. Weaviate was able to perform well within their required benchmark ranges.
• Reliable support: The company has an open culture of sharing what’s working and not working in their Slack channels for third-party developer tools. The Weaviate channel became a source of internal collaboration, problem solving, and best practice sharing with the Weaviate team. In addition, the company has peace-of-mind knowing they can access Weaviate’s 24/7 support team.