DataStax is updating its tools for building generative AI-based applications to ease and accelerate application development for enterprises, databases, and service providers.
One such tool is Langflow, which DataStax acquired in April, an open-source, web-based, no-code graphical user interface (GUI) that enables developers to visually prototype and iterate on LangChain flows to develop applications more quickly.
LangChain is a modular framework for Python and JavaScript that simplifies the development of applications powered by generative AI language models (LLMs).
The update to Langflow is a new version called Langflow 1.0, and it's the official open-source release that came after months of community feedback on previews, according to the company's chief product officer, Ed Anoff.
“Langflow 1.0 adds more flexible and modular components and capabilities to support more advanced Search Augmentation Generation (RAG) techniques and complex AI pipelines required for multi-agent architectures,” Anuff said, adding that Langflow's execution engine is now Turing complete.
Turing completeness or completeness is a term used in computer science to describe a programmable system that can perform or solve any computational problem.
The company said Langflow 1.0 also comes with LangSmith integration, which allows enterprise developers to monitor LLM-based applications and achieve observability over them.
A managed version of Langflow will also be available via DataStax in public preview.
“Details of the Astra DB environment will be available on Langflow and users can access Langflow via the Astra portal. It is free to use,” Anuff explained.
New features added to RAGStack 1.0
DataStax also released a new version of RAGStack, a curated stack of open-source software for implementing RAG in generative AI-based applications using Astra DB Serverless or Apache Cassandra as the vector store.
The new version, called RAGStack 1.0, includes new features such as Langflow, Knowledge Graph RAG, and ColBERT.
The Knowledge Graph RAG feature offers an alternative way of retrieving information using graph-based representations, which the company adds can be more accurate than just vector-based similarity search using Astra DB.
Other features include the introduction of Text2SQL and Text2CQL (Cassandra Query Language) to ingest any kind of data into generative AI flows for application development.
DataStax offers a separate unmanaged version of RAGStack 1.0 called Luna for RAGStack, but Anuff says the managed version offers more value to enterprises.
“RAGStack is based on open source components, and you can put all those projects together yourself. But we think it's really valuable for businesses to be able to test and integrate the stack so they can trust that it will deliver at scale the way they want it to,” the chief product officer explained.
The company has also partnered with several other companies, including Unstructured, to help developers extract, transform and store data in AstraDB to build generative AI-based applications.
“Our partnership with Unstructured will enable DataStax customers to use Unstructured's capabilities to extract and transform data in multiple formats, including HTML, PDF, CSV, PNG and PPTX, and convert it to JSON files for use in their AI initiatives,” said Matt Aslett, director of Ventana Research, ISG.
Other partnerships include collaborations with top embedding providers such as OpenAI, Hugging Face, Mistral AI and Nvidia.
