AI Factory Use Cases ==================== AI Factory is designed to help you build **Sovereign AI** solutions — where your **data**, **models**, and **AI-driven applications** operate under your governance, in your infrastructure, with your trusted content. These use cases highlight how AI Factory components work together to support **real-world, production-ready Sovereign AI applications** — across industries and solution patterns. For industry-specific solution examples, see: :ref:`Industry Solutions with AI Factory ` . ``---`` Retrieval-Augmented Generation (RAG) ------------------------------------ **RAG** is a foundational pattern for Sovereign AI — it lets you deliver **accurate, grounded responses** by combining **LLMs** with your enterprise data. What ^^^^ - Use AI models that **retrieve trusted content** from your Knowledge Bases, rather than relying only on opaque base models. - Make responses **auditable** and **explainable** — key for Sovereign AI. Why ^^^ - Reduce hallucinations. - Maintain compliance and traceability. - Ensure AI answers reflect your latest business knowledge. How ^^^ - Build :ref:`Create a Knowledge Base in Gen AI Builder ` from your Data Sources. - Implement tuned :ref:`Create a Retriever in Gen AI Builder ` for semantic search. - Power :ref:`Create an Assistant in Gen AI Builder ` and apps with RAG pipelines. Learn More ^^^^^^^^^^ - :ref:`Sovereign AI Explained ` - :ref:`Best practices for Hybrid Knowledge Bases ` ``---`` Document Intelligence Pipelines ------------------------------- Automate the transformation of documents into **structured, queryable knowledge** that remains **inside your governance boundary**. .. _what-1: What ^^^^ - Apply **OCR**, **parsing**, **summarization**, and **metadata extraction**. - Store results in :ref:`Create a Knowledge Base in Gen AI Builder ` or **Vector Engine**. .. _why-1: Why ^^^ - Keep sensitive documents in your infrastructure — no third-party API calls required. - Build explainable pipelines for compliance-heavy domains. .. _how-1: How ^^^ - Use :ref:`Pipelines and data preparation ` . - Leverage built-in :ref:`Pipelines reference - Preparers ` (OCR, parse PDF, parse HTML). - Index results for search or RAG. .. _learn-more-1: Learn More ^^^^^^^^^^ - :ref:`Sovereign AI Concepts ` - :ref:`Pipeline Getting Started ` - :ref:`Document Processing Examples ` ``---`` Conversational Assistants and Chatbots -------------------------------------- Build Assistants that run **within your infrastructure**, use **your models**, and rely on **your Knowledge Bases** — with complete visibility and control. .. _what-2: What ^^^^ - AI Assistants powered by **Model Serving** and your **RAG pipeline**. - Defined and governed with :ref:`Create a Ruleset in Gen AI Builder ` . .. _why-2: Why ^^^ - Avoid relying on public API services for LLM responses. - Control tone, style, compliance — essential for Sovereign AI. .. _how-2: How ^^^ - Build :ref:`Create an Assistant in Gen AI Builder ` in Agent Studio. - Govern behavior with :ref:`Rulesets (optional) ` . - Add :ref:`Create a Tool in Gen AI Builder ` to safely integrate internal systems. .. _learn-more-2: Learn More ^^^^^^^^^^ - Agent Studio in Hybrid Manager - :ref:`Sovereign AI Explained ` - :ref:`Getting Started - AI Factory 101 ` ``---`` Semantic Search Across Enterprise Content ----------------------------------------- Enable **natural language search** on your content — while ensuring embeddings and search data remain **inside your control**. .. _what-3: What ^^^^ - Vector-based search on **Knowledge Bases** or **Vector Engine**. - Supports text, hybrid structured/unstructured data, images. .. _why-3: Why ^^^ - Keep embeddings in your databases or object storage — not in third-party vector DBs. - Maintain auditability and security. .. _how-3: How ^^^ - Build :ref:`Create a Knowledge Base in Gen AI Builder ` . - Use :ref:`Create a Retriever in Gen AI Builder ` to expose search to Assistants or apps. - Use :ref:`Vector Engine (PGvector) ` for in-database search. .. _learn-more-3: Learn More ^^^^^^^^^^ - :ref:`Vector Engine Hub ` - :ref:`Knowledge Bases Reference Manual ` - :ref:`Building Production-Ready AI Features - AI Factory 201 ` ``---`` Real-Time Model Inference APIs ------------------------------ Deploy and serve models **inside your Kubernetes clusters**, under your governance — not through opaque third-party APIs. .. _what-4: What ^^^^ - Serve LLMs, embeddings, vision models via **KServe-based Model Serving**. - Models pulled from your :ref:`Model Library Reference Manual ` , deployed via :ref:`How-To Create an InferenceService for NVIDIA NIM Container ` . .. _why-4: Why ^^^ - Full control of model stack and model versions. - Run inference in **GPU-accelerated, air-gapped environments** if needed. .. _how-4: How ^^^ - Manage models in :ref:`Model Library Reference Manual ` . - Deploy with :ref:`Model Serving Reference Manual ` . - Secure endpoints with Hybrid Manager controls. .. _learn-more-4: Learn More ^^^^^^^^^^ - :ref:`Model Serving Concepts ` - :ref:`Sovereign AI Explained ` - :ref:`Advanced AI Factory Usage and Extensibility - AI Factory 301 ` ``---`` Automated AI-Powered ETL Pipelines ---------------------------------- Keep **AI data pipelines** inside your controlled environment — with no data leakage. .. _what-5: What ^^^^ - Use :ref:`Create a Structure in Gen AI Builder ` and Pipelines to run: - AI categorization - Document summarization - Metadata extraction - Data cleansing .. _why-5: Why ^^^ - Avoid sending sensitive data to cloud APIs. - Build reusable, explainable ETL components. .. _how-5: How ^^^ - Build :ref:`Create a Structure in Gen AI Builder ` . - Run as Tools in :ref:`Create an Assistant in Gen AI Builder ` or standalone Pipelines. - Monitor and govern pipeline execution via Hybrid Manager. .. _learn-more-5: Learn More ^^^^^^^^^^ - :ref:`Structures Reference Manual ` - :ref:`Advanced AI Factory Usage and Extensibility - AI Factory 301 ` ``---`` Summary ------- **Sovereign AI** use cases are about keeping **data, models, and AI applications under your governance**: - Build **RAG pipelines** grounded in your enterprise data - Automate document intelligence workflows - Deploy conversational Assistants in your infrastructure - Serve AI models via GPU-backed Model Serving - Run AI-powered ETL pipelines with no data leaving your environment - Power enterprise semantic search with full control of embeddings ``---`` **Next steps:** - Explore :ref:`Industry Solutions with AI Factory ` - Follow a :ref:`Learning Paths ` - Get started with :ref:`Gen AI How-To Guides ` ``---`` **AI Factory** gives you the foundation to build **trusted, governed, explainable Sovereign AI** — on your terms. → Start today with :ref:`Getting Started - AI Factory 101 ` , or deploy your first :ref:`Create an Assistant in Gen AI Builder ` .