Skip to main content

LLM Provider Integrations — BYOM (Bring Your Own Model)

NudgeBee uses flexible AI models — including modular SLMs, LLMs, and specialized agents — to power NuBi, the pre-built Cloud-Ops agents, root cause analysis, automated runbooks, and intelligent recommendations. This section guides you through connecting an LLM provider to your NudgeBee instance.

Do You Need This?​

  • Cloud SaaS / Enterprise users: A NudgeBee-managed LLM is available. You can skip this section unless you want to use BYOM (Bring Your Own Model) for more control over model selection or data handling.
  • Community (open-source) users: You need to configure your own LLM provider for AI features to work — this is the BYOM path. Choose from the options below.
info

Without an LLM connection, NudgeBee still provides monitoring, cost optimization, and alerting. The LLM unlocks NuBi and the full suite of AI-powered troubleshooting, natural language queries, agentic automation, and auto-generated runbooks.

Your Options — Flexible AI Models​

NudgeBee supports BYOM (Bring Your Own Model) with three categories of LLM providers:

CategoryProvidersBest For
Cloud Provider ServicesAWS Bedrock, Azure OpenAI, Google Vertex AI, Google Gemini, OpenAITeams with existing cloud contracts or preferred providers.
Self-Hosted / Open SourceOllama, HuggingFace, AWS SageMakerOrganizations requiring data privacy, air-gapped environments, or custom-trained models.
NudgeBee Models Enterprise CloudPre-trained NudgeBee AI models (nb-llm, nb-slm)Enterprise/Cloud users who want optimized, purpose-built models for Cloud Ops.

Supported LLM Providers​

  • AWS Bedrock is the default provider for LLM Server and RAG Server applications.

Choose from the following LLM providers to integrate with your NudgeBee applications:

Cloud Provider Services​

  • AWS - Amazon Web Services integration options including Bedrock and SageMaker
  • Azure - Microsoft Azure integration options including Azure OpenAI Service
  • Google - Google Cloud Platform integration options including Gemini and VertexAI
  • OpenAI - OpenAI API integration for GPT-5, GPT-4o, GPT-4, and Embeddings models

Open Source & Self-Hosted Options​

  • Hugging Face - Integration with Hugging Face's model repository and inference APIs
  • Ollama - Integration with self-hosted Ollama deployments

NudgeBee Models Enterprise Cloud​

Enterprise feature

NudgeBee's pre-trained nb-llm / nb-slm / nb-text-embeddings models are part of the Enterprise and Cloud editions and are downloaded from the licensed registry.nudgebee.com registry. Community (open-source) users should connect their own model instead — see the Open Source & Self-Hosted Options above (Ollama, Hugging Face) or any BYOM provider. See Editions for details.

NudgeBee provides pre-trained AI models that can be downloaded and deployed on supported platforms (applicable for licensed on-premises or self-hosted Enterprise environments):

NudgeBee AI/LLM Models

  • Download pre-trained AI models from the NudgeBee platform using the following commands (requires an Enterprise license key):

    SLM​
      curl --location 'https://registry.nudgebee.com/downloads/models/nb-slm' --header 'Authorization: Bearer <license_key>'
    LLM​
      curl --location 'https://registry.nudgebee.com/downloads/models/nb-llm' --header 'Authorization: Bearer <license_key>'

  • Optimized for high-performance inference in various AI-driven applications.

Models Used for Retrieval-Augmented Generation (RAG)​

RAG models enhance information retrieval by generating vector embeddings and enabling efficient similarity searches:

  1. nb-text-embeddings
    • Generates vector embeddings for text data.
    • Powers the RAG Server for efficient similarity searches and context retrieval.

Models Used for Agents (LLM Server)​

The LLM Server powers intelligent agents that specialize in reasoning, planning, and query generation:

  1. nb-llm

    • Functions as the primary reasoning and planning model.
    • Handles complex query processing, decision-making, and response generation.

  2. nb-slm

    • Designed for task-specific agents, improving modular AI functionality.