ServiceNow has partnered with NVIDIA to introduce Apriel Nemotron 15B, a new 15-billion-parameter reasoning model designed to accelerate Artificial Intelligence adoption in enterprise environments. Announced at ServiceNow’s Knowledge 2025 event with NVIDIA CEO Jensen Huang and ServiceNow CEO Bill McDermott, the compact model aims to streamline real-time responses and automate complex workflows across IT, HR, and customer service teams globally.
The model was engineered using NVIDIA’s NeMo platform, the open Llama Nemotron Post-Training Dataset, and ServiceNow’s domain-specific data, then trained on NVIDIA DGX Cloud hosted on Amazon Web Services. Apriel Nemotron 15B distinguishes itself with its smaller size compared to much larger general-purpose models, resulting in faster response times and reduced costs while maintaining robust enterprise-grade intelligence. The design is intended to power thousands of simultaneous enterprise workflows and AI agents with high efficiency and scalability.
ServiceNow and NVIDIA have also integrated a data flywheel architecture, leveraging ServiceNow’s Workflow Data Fabric in concert with NeMo microservices such as the NeMo Customizer and Evaluator. This closed-loop system allows for personalized, continuously improving responses based on workflow data, with strict guardrails to ensure data security and compliance. Demonstrations at the event highlighted deployments with major enterprises like AstraZeneca, where these AI agents are expected to yield significant productivity improvements. The launch is positioned as a key advancement in enterprise Artificial Intelligence, moving from static models to dynamic, evolving systems and reinforcing the ongoing ServiceNow-NVIDIA collaboration to push agentic AI forward.
The rollout of ServiceNow AI Agents powered by Apriel Nemotron 15B will commence after Knowledge 2025, supporting the Now LLM services and enhancing the company’s advanced agentic AI offerings for enterprise customers.
