Cisco, NVIDIA, and VAST Data develop architecture for agentic AI

"The project, called Cisco Secure AI Factory with NVIDIA, combines Cisco's network and server technology with NVIDIA's accelerated computing power and AI software and VAST Data's data platform technology. In this way, Cisco is continuing to build on the Cisco AI PODs offering together with its partners. These are modular building blocks that can process data faster and make it available for AI applications. With the addition of the VAST InsightEngine, these PODs can perform retrieval-augmented generation, or RAG, more efficiently."

"This technique combines large language models with current business data, making answers more reliable, specific, and immediately usable for business applications. This reduces the time needed to extract information from business data and make it available to AI models. The architecture must ensure that AI agents have access to the right information in near real time, without delays caused by data bottlenecks. At the same time, attention is paid to security and governance, so that sensitive data remains protected and companies comply with compliance requirements."

Cisco, NVIDIA and VAST Data jointly deliver a modular architecture—Cisco Secure AI Factory with NVIDIA—integrating Cisco network and server technology, NVIDIA accelerated computing and AI software, and VAST Data's data platform. The design builds on Cisco AI PODs as high-performance building blocks that ingest and process data rapidly, with VAST InsightEngine enabling efficient retrieval-augmented generation (RAG). RAG fuses large language models with up-to-date business data to produce more reliable, specific, and immediately usable outputs for enterprise workflows. The architecture emphasizes near-real-time data access, scalable support for multiple AI workloads, and integrated security and governance to protect sensitive data and ensure compliance. The first RAG-accelerated AI POD is now available.

Read at www.techzine.eu