DeeperML

ServiceNow AI has launched Apriel-5B, a new family of small language models (SLMs) designed to rival larger language models (LLMs) while requiring fewer resources. This release addresses the growing challenge of deploying large-scale models, which often demand significant infrastructure and incur high operational costs, making them inaccessible to many organizations. Apriel-5B, with its 4.8 billion parameters, aims to bridge the gap between capability and deployability, offering competitive performance on instruction-following and reasoning tasks while being deployable on modest hardware. The Apriel family includes Apriel-5B-Base, a pre-trained model for further tuning, and Apriel-5B-Instruct, an instruction-tuned version for chat, reasoning, and task completion. Both models are released under the MIT license to encourage open experimentation and broader adoption across various sectors.

ServiceNow is also heavily invested in integrating AI across its operations, including significant R&D and engineering efforts in India. According to Paul Smith, president of global customer and field operations at ServiceNow, 85% of ServiceNow India's resources are dedicated to R&D and engineering. This reflects India's crucial role in building the future of enterprise automation, as the engineering team has grown at an average annual rate of 25% over the last three years. These efforts are paying off, with the company currently running over 200 agentic use cases in production, generating $325 million in annual savings. This includes improving internal processes such as finance help desk turnaround times from four days to just eight seconds using ServiceNow agents, allowing redeployment of staff to more strategic roles.

Furthermore, ServiceNow is leveraging AI to enhance its AIOps capabilities, building an AIOps chatbot with Amazon Q Business custom plugins. This chatbot integrates multiple third-party applications into a unified, natural language-driven interface. By using OpenAPI schemas, the chatbot can interact with AWS infrastructure, enabling users to query data and perform actions using natural language prompts. For instance, users can ask the chatbot to check Amazon S3 bucket access settings or close open ports on Amazon EC2 instances. This streamlines operational tasks and simplifies complex cloud operations, enabling IT teams to manage infrastructure through natural language interactions and reducing manual intervention.


References :

• analyticsindiamag.com: 85% of ServiceNow India Resources are in R&D & Engineering, says President Paul Smith
• learn.aisingapore.org: Building an AIOps chatbot with Amazon Q Business custom plugins
• www.marktechpost.com: Small Models, Big Impact: ServiceNow AI Releases Apriel-5B to Outperform Larger LLMs with Fewer Resources
• MarkTechPost: Small Models, Big Impact: ServiceNow AI Releases Apriel-5B to Outperform Larger LLMs with Fewer Resources
• Ken Yeung: ServiceNow Expands Its AI Ambitions With New Apriel Small Language Models

Classification:

• HashTags: #ServiceNowAI #SmallLLM #AIops
• Company: ServiceNow
• Target: Enterprises
• Product: Apriel
• Feature: Apriel-5B
• Type: AI
• Severity: Informative