Cisco has added VAST Data’s InsightEngine to its NVIDIA-integrated AI system to enable retrieval-augmented generation (RAG) and agentic AI applications.

The announcement is significant on several fronts. First, it adds capabilities to the Cisco Secure AI Factory platform announced with NVIDIA at Cisco Live event in June. That platform is the linchpin of Cisco’s AI networking strategy, which emerged from the Cisco reorganization last year that gave more power to Jeetu Patel, Cisco’s president and chief product officer. That move has resulted in new energy on the product side—particularly when it comes to AI.

Cisco’s announcement also draws upon the existing VAST-NVIDIA partnership.

VAST’s InsightEngine, which is geared to AI processing, can access NVIDIA’s NeMo Retriever microservices, which in turn are part of the NVIDIA Enterprise AI platform. When data comes into the VAST Data platform, InsightEngine links to the NVIDIA microservices, which in turn create vector embeddings that are then used to update the vector index in the VAST DataBase. According to the vendors, this process ensures that any new file, object, table, or streaming data is instantly ready for AI retrieval and inference.

Cisco's news highlights the importance of RAG in enterprise AI. RAG is the process of integrating external data to provide context for large language models (LLMs) and small language models (SLMs), theoretically improving the accuracy of responses. RAG is also useful for keeping up with data that changes dynamically.

Building AI with Cisco PODs

The integrated Cisco-NVIDIA-VAST solution is available on Cisco’s AI PODs, which are blocks of storage, compute, and networking hardware and software geared to AI inferencing. The PODs are based on NVIDIA’s AI Data Platform reference design, which blueprints the infrastructure for running NVIDIA Enterprise AI software.

These AI PODs also come with Cisco UCS servers embedded with NVIDIA RTX PRO 6000 Blackwell Server Edition GPUs. Hence, the Cisco AI POD provides a full platform for making raw data ready for RAG.

But there’s more. Cisco is also intent on using this solution to deliver updated data to AI agents. Jeremy Foster, SVP and GM, Cisco Compute, said in the press release:

“Agentic AI has the potential to unlock the value of AI for enterprises around the world. Moving beyond chatbots to agents that can help solve true business challenges is revolutionary, but only if enterprises can effectively leverage the right data at the right times. Cisco, NVIDIA and VAST are working together to give customers a simple path to unlocking the value of their data.”

Up to now, many organizations have labored to deliver optimized data to RAG pipelines. By assembling this integrated solution, Cisco is offering a comprehensive approach toward RAG and inference. It also demonstrates Cisco's intent to leverage its value in networking and security to address enterprise AI requirements.