DataStax unveils major updates to AI platform at RAG++ event

DataStax has announced significant updates to its AI platform at the RAG++ event in New York City, aimed at enhancing the performance and usability of generative AI (GenAI) applications. The company has introduced new features designed to simplify data preparation and ingestion and improve query relevancy, response times, and throughput.

One of the key updates is the native integration with Unstructured.io, which facilitates the transformation of various data formats into a state that is suitable for retrieval-augmented generation (RAG). Developers often struggle with the ingestion of large documents, and this integration allows them to import and process multiple PDF files efficiently. By using DataStax Vectorize, these documents can be chunked into smaller segments, making it easier to generate vector embeddings for improved query accuracy. This integration supports more file types and streamlines the data handling process, thus optimising application resource utilisation.

"Data preparation is a common issue for developers as they build their GenAI apps. They need to ingest, process, and chunk more data to ensure applications are delivering accurate, relevant query responses," said Brian Raymond, CEO of Unstructured. "With our new, native integration with Langflow and Astra DB, we're allowing AI developers to easily import and process unstructured data like PDFs, emails, and more. This enhanced capability sharpens query results and centralises unstructured data handling within DataStax's AI PaaS."

Another notable enhancement is the integration of Glean, a tool that enables seamless access to data stored in Astra DB. This feature allows Glean to directly analyse the data, providing relevant and accurate query responses. Users can also leverage Glean's indexing capabilities to enrich the context of their operations, making decisions based on real-time data insights.

"DataStax Langflow makes developing AI apps easy," noted Arvind Jain, CEO of Glean. "Now with Glean built-in, developers can connect to all their important corporate data sources and build custom AI experiences that help their company automate work with AI. DataStax plus Glean will enable both structured and unstructured data to feed AI workflows."

In addition to these integrations, DataStax has introduced a public preview of the DataStax Langflow API. This new API allows developers to build and host their GenAI applications with a simple HTTP call, eliminating the need for self-hosting the applications. The API includes JavaScript and Python code snippets that can be easily integrated into existing projects, providing a faster path to production.

"As developers move beyond the ideation and experimental phase that has characterised the past year or so, they're looking to deploy GenAI applications into production with ease," said Ed Anuff, Chief Product Officer at DataStax. "The DataStax AI PaaS offers users the ability to quickly build, iterate, and deploy applications with speed, at scale. It's a field-proven platform that enables some of the largest global companies to leverage their data to power production-ready GenAI applications and deliver new internal and customer-facing experiences to the market."

Brendon Geils, Founder of Athena Intelligence, also weighed in on the new features. "We rely on Langflow's underlying infrastructure to provide a robust environment for our customers to build and deploy their own GenAI applications with ease," he explained. "Our users build custom flows using our UI, and we rely on the orchestration and automation provided by Langflow under the hood to make that happen. The Langflow API will provide more flexibility and stability on our platform, providing data analysts with the seamless experience they need to deploy and scale purpose-built AI applications for their day-to-day workflows."

DataStax's latest updates aim to support developers in simplifying the GenAI application development process, enhancing performance metrics, and integrating more effectively with other technologies.