Alfresco AI Framework

The Alfresco AI Framework is a robust framework designed to integrate AI capabilities into Alfresco, leveraging Java and Spring AI. It provides a suite of tools and services to process, analyze, and enhance document content in Alfresco using AI and machine learning models.

Note: This project uses the Spring AI SNAPSHOT version, as a final RELEASE is not yet available.

Key Features

• AI-Driven Document Processing: Enable Retrieval-Augmented Generation (RAG) chat functionality within the Alfresco repository.
• Modular Architecture: Built on a Spring Boot-based architecture, offering flexibility in deployment and scalability.
• Seamless Integration: Easily extend Alfresco Content Services with AI capabilities.

Getting Started

Prerequisites

• Java 17: Required for building and running the services
• Maven: For building the backend services
• Angular 18: For developing and running the UI
• Node.js 18: Required for building the UI
• Docker: For containerized deployment

Architecture

Projects

• ai-rag-framework:
  A REST API built on top of Spring AI for ingesting documents into a Generative AI (GenAI) model and providing a Retrieval-Augmented Generation (RAG) chat service.

  • Requirements:
    • A local Large Language Model (LLM) powered by Ollama
    • Elasticsearch as the vector database

• alfresco-ai-sync:
  A service built on top of Alfresco Java SDK that retrieves documents from the Alfresco Repository and ingests them into the vector database through the ai-rag-framework API.

  • Requirements:
    • A running Alfresco Repository
    • The ai-rag-framework service

• alfresco-ai-ui:
  A user interface built on top of Alfresco ADF for interacting with the RAG chat service provided by ai-rag-framework.

  • Requirements:
    • The ai-rag-framework service must be running

• alfresco-docker: Container-oriented Alfresco Community 23.3 deployment

Laboratories

This series of tutorials will guide you through the key features of the project, including data ingestion, chat integration, and the overall operation of the system.

Lab 1: Ingestion Pipeline

In this lab, you'll learn how to populate the Vector Database (Elasticsearch) with selected content from the Knowledge Base stored in Alfresco. This involves extracting vectors from content using the Embedding module nomic-embed-text via Ollama.

Begin the lab by following Lab 1: Ingestion Pipeline.

Lab 2: Chat Functionality

This lab focuses on enabling chat functionality with the LLM qwen2.5 via Ollama, using Alfresco UI apps like Share and ADF. The process includes transforming the user's prompt into vectors using the Embedding module nomic-embed-text via Ollama, then searching for relevant content in the Vector Database (Elasticsearch). The retrieved text is used to provide context to the LLM, which helps generate more accurate responses.

Start this lab by following Lab 2: Chat Functionality.

Lab 3: Running All Components Together

In this lab, you'll integrate all the components (ingestion and chat functionality) with a live Alfresco Repository. The system will automatically update the Vector Database whenever there are changes in the repository, eliminating the need for manual intervention.

You can begin this lab by following Lab 3: Running All Components Together.

License

This project is licensed under the Apache License 2.0. See the LICENSE file for details.

Acknowledgments

Special thanks to the Alfresco and Hyland teams for their continuous support and contributions to open-source initiatives in the content management and AI domains.