At VAST Forward 2026, VAST Data unveiled two new computing services — VAST Data PolicyEngine and VAST Data TuningEngine — designed to advance the capabilities of its AI Operating System and support organisations scaling mission-critical AI deployments.
The two services are engineered to work in tandem, enabling AI systems that are governed, explainable and continuously learning. PolicyEngine is focused on governing agentic activity, while TuningEngine manages model tuning and reinforcement learning workflows. Together, they create automated learning loops designed to remain aligned with organisational policies and expectations.
“Just as people are always learning, so should tomorrow’s applications,” said Jeff Denworth, co-founder at VAST Data. “With the introduction of PolicyEngine and TuningEngine, the VAST AI Operating System has become a thinking machine that customers can deploy wherever they compute – a machine that safeguards every interaction and learns from every outcome, bringing the power of AI within reach of every organization.”
Strengthening governance in agentic AI
As AI agents increasingly access enterprise data and generate new information — from model outputs to agent-to-agent communications — governance has become a critical requirement. Without granular controls and auditability, risks such as data leakage and policy violations increase.
VAST’s PolicyEngine addresses this challenge through inline policy enforcement that governs agent access to shared memory, tools, knowledge bases and other agents. The system applies fine-grained, explicit permissions and AI-derived contextual controls before actions are executed. It also maintains tamper-proof logs and traceability, reinforcing a zero-trust operating model designed to ensure that agent decisions remain observable, explainable and auditable.
Enabling continuous model improvement
Complementing PolicyEngine, the TuningEngine extends VAST’s AgentEngine — the AI OS’s serverless agentic runtime — by introducing structured learning loops. While AgentEngine supports multi-agent orchestration and model deployment, TuningEngine captures performance data from agent workflows and uses curated feedback to continuously improve models.
Using techniques such as LoRA fine tuning, supervised fine tuning and reinforcement learning, TuningEngine automates data ingestion, candidate model generation and benchmarking within the VAST AI OS. Approved models can then be deployed manually or automatically, initiating new cycles of improvement based on future interactions.
The TuningEngine will also integrate with NVIDIA’s NeMo Data Designer to support training and fine-tuning of NVIDIA Nemotron open models, expanding VAST’s collaboration with NVIDIA.
With the introduction of PolicyEngine and TuningEngine, VAST Data said its AI OS now enables a closed operational loop that observes, reasons, acts, evaluates and improves — while embedding governance and security at every stage.
The new services are expected to be available by the end of 2026.