September 2023 Summaries

The text discusses the use of Generative AI in improving productivity and efficiency within industries facing challenges such as an aging population, resource constraints, labor shortages, and regulatory issues. SkyPoint AI, a SaaS provider of data, analytics, and artificial intelligence services, is highlighted for its innovative use of AI and DataStax Astra DB, a vector database built on Apache Cassandra®. Key takeaways from Tisson Mathew's approach include mastering data integration, using knowledge graphs, incorporating AI for decision support, automating processes, prioritizing user-centric design, and ensuring affordability.

Generative AI (Gen AI) applications require fresh, real-time data from various sources to build accurate context and avoid hallucinations. To address this challenge, DataStax Astra DB added vector search capabilities. Now, the introduction of LangStream, an open-source project, combines event-based architectures with Gen AI technologies. LangStream makes it easy to coordinate data from diverse sources for high-quality prompts for large language models (LLMs), enabling scalable, production-ready, real-world AI applications on a broad range of data types. The integration of LangStream and Astra DB is particularly useful for constructing high-quality prompts, making Gen AI interactions more intuitive and effective.

DataStax has added vector search capabilities to its Astra DB, a database-as-a-service built on open source Apache Cassandra, to help developers create AI applications. The new JSON API for Astra DB enables JavaScript developers to access the most powerful vector database in the world through a simple API. This integration is fully compatible with MongooseJS, a popular framework for object modeling on top of document databases. With this addition, JavaScript developers can now easily build enterprise-grade Gen AI applications using familiar paradigms and frameworks. The JSON API is currently in public preview and available to anyone who wants to try it out.

Vector search technology is being utilized to enhance a NodeJS application with generative AI capabilities by leveraging DataStax Astra DB and its vector search support through the stargate-mongoose driver and Stargate's JSON API. An engineer developed a NodeJS app called Photography-Site, designed to organize and showcase photographic work, integrating Astra DB's vector search for both text and image similarity searches. This involves using the OpenAI embedding API for text embeddings and Google MediaPipe for image embeddings, enabling users to search for photos by description or visual similarity. The application is built on the Express framework, employing Mongoose as the driver to access data in Astra DB, and it features functionalities such as image browsing, adding images, and searching with vector similarity. The stargate-mongoose driver, an alternative to the MongoDB driver, offers JavaScript developers a JSON-oriented data model experience with Cassandra's scalability, marking a significant advancement in Cassandra's evolution. The demo app, available in the stargate-mongoose-sample-apps repository, showcases the integration and capabilities of vector search in serverless Astra DB, providing a practical example of combining AI models with database technologies.