At ServiceNow’s flagship Knowledge 2025 event in Las Vegas, ServiceNow and NVIDIA unveiled Apriel Nemotron 15B, a domain-specific large language model (LLM) tailored to enterprise use cases. Designed to enhance the intelligence of real-time AI agents, this 15-billion-parameter model is integrated directly into ServiceNow’s platform and built on NVIDIA’s NeMo framework.

What sets Apriel Nemotron 15B apart is its enterprise-grade reasoning, enabling it to follow business logic, weigh competing priorities, and respond to real-time conditions with precision. Despite being smaller than many general-purpose models, it’s optimized to tackle complex workflows in departments such as IT service management, HR, and customer operations—all with greater efficiency and explainability.

The model forms the foundation of ServiceNow’s upcoming Now LLM services and intelligent agents. These agents are built with what ServiceNow describes as a “data flywheel” architecture, combining real-time workflow data with NVIDIA microservices to constantly refine performance and personalization.

Importantly, Apriel Nemotron 15B isn’t just theoretical, it’s already showing real-world results. AstraZeneca, for example, is piloting ServiceNow's generative AI agents and reports potential productivity savings of up to 90,000 staff hours annually.

With its seamless integration into ServiceNow’s platform, the Apriel Nemotron 15B eliminates the need for overhauling legacy infrastructure, making it accessible to enterprises of all sizes. It represents a shift toward smaller, purpose-built models that prioritize accuracy, speed, and alignment with business needs.