VAST Data, the AI Operating System company, announced today that it is delivering a complete data and compute platform that enables AI intelligence to continuously evolve. The VAST AI OS, combined with NVIDIA AI Enterprise, which includes NeMo microservices that power a data flywheel for continuous model improvement, creates a unified environment where AI pipelines can constantly learn, adapt, and improve. This reference workflow provides a self-optimising foundation for scalable AI, laying the groundwork for billions of intelligent agents to fine-tune and evolve from their data and experiences.
This solution provides enterprises with a converged software platform for data management, database services, and AI compute orchestration. Additionally, VAST AI OS AgentEngine uniquely shares feedback by providing the critical capability to map the intricate web of agent-data interactions through production logs. This granular traceability allows the flywheel to dissect these multi-step interactions, accurately identifying which specific elements require adjustment to enhance outcomes, accelerating model performance and accuracy at scale.
“AI-powered businesses need thinking machines designed for a future where billions of AI agents learn from their own experiences, fine-tune in real time, and create new possibilities through collaboration,” said Jeff Denworth, Co-Founder of VAST Data. “By unifying NVIDIA’s AI software and hardware technologies within the core of the VAST AI Operating System, we are giving customers the foundation to operationalise continuous improvements in AI intelligence at scale, with the security, governance, and service delivery tools required to manage these intelligent agents and the data they rely upon.”
As AI moves from isolated projects to always-on infrastructure, businesses need systems that evolve in real time with every data point while addressing new security and governance challenges around fine-tuned models, agent interactions, and decentralised data pipelines. This collaboration makes it possible to run continuous, automated AI pipelines – from ingestion to inference to retraining – all managed within the VAST AI Operating System.
Among the first to embrace this strategy is CACEIS, one of Europe’s largest asset servicing firms. In collaboration with VAST and NVIDIA, CACEIS is exploring a real-time AI platform concept designed to securely capture, transcribe, and analyse 100% of client meetings. The vision is for the system to instantly generate meeting minutes, surface actionable insights, and deliver anonymised trend data — all seamlessly integrated into their sovereign CRM. With an end-to-end security model at its foundation, the platform is being designed to safeguard client privacy and data integrity while continuously evolving through every interaction.
“AI will be a game-changer, highlighting trends in current needs by analysing meeting reports so we can better serve clients,” said Arnaud Misset, Chief Digital Officer, CACEIS.
Using VAST’s AgentEngine that leverages the NVIDIA AI-Q Blueprint, CACEIS is developing a platform proof of concept that would enable AI agents to assist relationship managers in real time and help uncover new business opportunities. Built by NVIDIA NeMo microservices and the NVIDIA data flywheel blueprint, the envisioned CACEIS AI factory would continuously capture data and insights from every customer interaction. These feedback loops are intended to drive ongoing model refinement and training, allowing the system to improve and adapt with each meeting. As AI agents learn from one another and from human counterparts, this concept sets the stage for new ideas, collective intelligence, and enterprise-wide knowledge sharing to take shape.
This collaboration signals a broader shift in enterprise AI, from one-size-fits-all models to dynamic ecosystems of intelligent agents that continuously fine-tune, collaborate, and generate new ideas from their own data and interactions. Managing these agents at scale requires fine-grained security, governance, and access controls to ensure they operate safely and within defined boundaries. It also demands scalable, dynamic infrastructure capable of handling the varied and unpredictable demands of AI agents as they interact with diverse data tools and services. The VAST AI Operating System provides this real-time datainfrastructure, along with compute orchestration, QoS tools that enforce fairness as different agents run within the environment, and a security framework to refine intelligence and operationalise AI innovation in a secure, scalable, and adaptive system.
“Data flywheels leverage each interaction with an AI agent to continuously improve system intelligence and value,” said Kari Briski, Vice President of Generative AI Software at NVIDIA. “CACEIS is an exemplary pioneer with a vision of building an agentic AI data flywheel with NVIDIA and VAST to supercharge productivity for financial services in Europe.”