Nexla Inc., a data integration platform for generative artificial intelligence, today announced the integration of Nvidia Corp.’s NIM microservices and accelerated computing to enable faster, simpler and more cost-efficient development of production-grade retrieval-augmented generation pipelines.

Production-grade RAG pipelines for enterprise AI applications combine advanced data retrieval and generative AI capabilities to create highly accurate, context-aware responses. The pipelines integrate structured and unstructured data sources to allow enterprises to process large datasets efficiently and deliver actionable insights in real-world AI systems, such as co-pilots and intelligent workflows.

“Internal data is key to getting value out of AI pipelines, delivering relevant responses and building AI solutions that meet the needs of enterprises,” said Jeremy Krinitt, senior developer relations management at Nvidia.

Nexla’s visual design for data integration and unified retrieval across structured and unstructured data, along with composable RAG pipelines, accelerates the building of production-grade generative AI. Nvidia NIM provides hardware acceleration and optimized microservices across key components, including multimodal ingestion, parsing, reranking and inferencing, reducing run-time latency and overall costs.

Through its collaboration with Nvidia, Nexla is aiming to help businesses with their generative AI strategies by supporting a smooth transition from demo to production-grade AI. Nexla’s data integration and workflow management for generative AI with NIM helps equip enterprises with the tools they need to leverage and scale vast amounts of unstructured and structured data for generative AI.

“Our integration tools with Nvidia’s hardware acceleration is an ideal fit for any organization looking to accelerate its journey toward the real-world deployment of AI systems,” said Nexla Chief Executive Saket Saurabh. “The integration of these two powerful engines can result in tremendous gains in productivity, performance and cost-efficiency for our customers, from proof of concept through production.”

Key benefits of the integration include document ingestion and accelerated embedding and retrieval. Nexla connectors enable powerful RAG processes by interpreting text, tables, charts and images to efficiently manage large volumes of structured and unstructured data, with Nvidia NIM enhancing the capability by accelerating document parsing and embedding generation, transforming data into vector embeddings optimized for RAG workflows.

Nexla’s platform also provides unified retrieval and scalable RAG workflows that leverage a metadata-powered approach to identify relevant datasets across different systems such as vector stores, application programming interfaces and databases. Through the use of GPU-powered containers, Nexla reduces the need for traditional non-GPU containers, provoding faster and more cost-efficient processing.

Image: SiliconANGLE/Ideogram