{"slug":"nexthink-fine-tunes-llms-on-amazon-sagemaker-for-its-nexthink-assist-ai-agent","url":"https://findausecase.com/use-cases/nexthink-fine-tunes-llms-on-amazon-sagemaker-for-its-nexthink-assist-ai-agent","title":"Nexthink fine-tunes LLMs on Amazon SageMaker for its Nexthink Assist AI agent","description":"Swiss software company Nexthink moved its NL2NQL tool, which translates natural-language questions into Nexthink Query Language queries for the Nexthink Assist agent, from prompt engineering and RAG to fine-tuned LLMs on Amazon SageMaker. The source states fine-tuning gave higher accuracy (+30%) and lower token usage (-80%), and describes an LLM-as-a-judge labeling pipeline using Claude 3.5 Sonnet on Bedrock.","company":"Nexthink","industry":"Technology & Software","country":"Switzerland","aiCapabilities":["Large Language Models","Natural Language Processing"],"technology":["Amazon SageMaker","Amazon Bedrock","Anthropic Claude 3.5 Sonnet"],"deployment":"Unknown","problemStatement":"Nexthink's platform had traditionally been reserved for expert IT power users, while other IT teams, HR and employees wanted the same insights. Its specialised Nexthink Query Language (NQL) created a barrier for non-technical users.","solutionApproach":"Nexthink Assist is an AI agent that lets users ask questions in plain language, powered by a tool called NL2NQL that translates questions into NQL queries. Early versions used prompt engineering, then retrieval-augmented generation with an Amazon OpenSearch vector database, and then fine-tuning of a 7 billion parameter Llama model with LoRA on Amazon SageMaker. Nexthink built a golden dataset and scaled labelling with an LLM-as-a-judge pipeline using Anthropic Claude 3.5 Sonnet on Amazon Bedrock, with humans periodically reviewing samples.","businessValue":"The source reports that fine-tuning gave a 30% improvement in query accuracy, an 80% reduction in token usage and support across the entire data model, and says it also lowered inference costs.","evidence":{"band":"high"},"sourceUrl":"https://aws.amazon.com/blogs/alps/how-nexthink-builds-enterprise-ai-agents-using-fine-tuned-llms-with-amazon-sagemaker","dates":{"publishedAt":"2026-11-15T05:48:57.026Z","publishedAtSource":"pipeline","updatedAt":"2026-11-15T05:48:57.026Z"},"license":"Open for reading and citing with a link to https://findausecase.com/use-cases/nexthink-fine-tunes-llms-on-amazon-sagemaker-for-its-nexthink-assist-ai-agent. Bulk republication requires permission."}