DataStax 'Data API' connects validation feeds to generative AI

Software Connections. When we use enterprise computer systems at their most efficient level, we tend to connect applications to other applications, other data services, other operating systems, and application components. If you've ever experienced the Internet, you'll understand why interconnectivity is important when working with applications, whether on a desktop computer or a smartphone.

Software applications themselves often use what has become a fairly standard approach to communication known as an Application Programming Interface (API). This is a way of creating a connection and channel (some would say "glue") between applications in a prescribed way (defined according to a consistent syntax) that allows software engineers to plug in all the necessary plugs - not at all like the stereotypical image of an old-fashioned telephone clerk plugging in cables at a PBX.

Real-time vector database company DataStax is trying to plug all the right wires into its approach to building generative artificial intelligence applications. This month, the company unveiled a new "Data API" (formerly known as JSON API, an acronym standing for JavaScript Object Notion, an open standard file format and data interchange format). The technology is intended to serve as a universal API for Retrieval Augmented Generation (RAG) development that provides all the data and a complete RAG stack for building generative AI applications with high relevance and low latency.

What is extended generation (RAG)?

The key is to create generative AI projects that have a significant connection to external sources of information for additional relevance, timeliness and accuracy. RAG extends the scope and application of so-called Large Language Models (LLMs) by connecting LLMs to external sources of validated, certified, or at least recognized knowledge data. This approach to generative AI augmentation was originally conceived, proposed and documented by parent company Facebook Meta; in the world of unbiased non-allusive generative AI, RAG is the lord and master of the manor these days.

In addition, the Data API product will include an updated developer interface for Astra DB, the company's vector database for building production-grade AI applications.

"Astra DB is ideal for JavaScript and Python developers, simplifying vector search and large-scale data management by hiding the power of Apache Cassandra behind a convenient yet powerful API," said Ed Anuff, Chief Product Officer at DataStax. "This release redefines the way software engineers build generative AI applications, offering a streamlined interface that simplifies and accelerates the development process for AI engineers."

Taking a closer look at how the product works, Anuff explains that the new API and vector data experience makes the petabyte scale of the open source Apache Cassandra database available to JavaScript and Python developers in a more intuitive way for AI development. By prioritizing ease of use, DataStax claims that now using the JVector search engine, we can get 20% higher relevance, 9x higher throughput, and up to 74x faster response times than other vector databases. In an effort to simplify the development process, the technology now features a dashboard, more efficient data loading (because data entry is always a headache), as well as data exploration tools and integration with other AI and machine learning (ML) frameworks.

Ready-made artificial intelligence API

"Software application developers can use the Data API to create a turnkey AI ecosystem that simplifies integration with major generative AI partners such as LangChain, OpenAI, Vercel, Google's Vertex AI, AWS, Azure and major platforms, while supporting a wide range of security and compliance standards," Anuff and his team confirmed. Any developer can now support advanced RAG techniques such as FLARE [Forward Looking Active REtrieval Augmented Generation] and ReAct, which must synthesize multiple responses while meeting latency SLAs."

Looking ahead, we are talking about generative AI with RAG that is capable of doing more and being more accurate. We're also talking about additional techniques overlaid on the RAG that make it more capable of knowing "when" and "if" to look for additional information in the future. In other words, we're talking about making AI more intelligent.

From what we're seeing with DataStax - and the usability factor is nice enough, but let's look deeper - this is the point at which we start to move generative AI applications forward with support for the vector databases they rely on as their lifeblood, and make them very flexible in deploying real-time data. Astra DB's experience allows for instant requesting real-time data updates for both vector and non-vector data, so let's remember where we came from.

Would a phone exchange with AI APIs now work seamlessly (that's a word technology vendors are so fond of using) and provide a perfect connection every time, with no line crossings and crystal clear reception from start to finish? Only a fool would bet on that level of perfection, even with artificial intelligence included.... hold, please, subscriber.