search

E.D.D.I Documentation

Multi-Agent Orchestration Middleware for Conversational AI. Coordinates multiple AI agents, business systems, and conversation flows. Developed in Java, powered by Quarkus, provided with Docker, and o

E.D.D.I (Enhanced Dialog Driven Interface) is a multi-agent orchestration middleware that coordinates between users, AI agents (LLMs), and business systems. It provides intelligent routing, conversation management, and API orchestration for building sophisticated AI-powered applications.

What EDDI Does:

  • Orchestrates Multiple AI Agents: Route conversations to different LLMs (OpenAI, Claude, Gemini, Ollama) based on context and rules

  • Coordinates Business Logic: Integrate AI agents with your APIs, databases, and services

  • Manages Conversations: Maintain stateful, context-aware conversations across multiple agents

  • Controls Agent Behavior: Define when and how agents are invoked through configurable rules

Developed in Java using Quarkus, it is lean, RESTful, scalable, and cloud-native. It comes as Docker container and can be orchestrated with Kubernetes or Openshift. The Docker image has been certified by IBM/Red Hat.

Latest stable version: 5.6.0

License: Apache License 2.0

Project website: herearrow-up-right

Codacy Badgearrow-up-right CircleCIarrow-up-right

alt text

hashtag
Intro

E.D.D.I is a multi-agent orchestration middleware for conversational AI systems. It sits between your application and multiple AI agents (LLMs, APIs, services), intelligently routing requests, coordinating responses, and maintaining conversation state.

EDDI as Orchestration Middleware:

  • Agent Coordination: Manage multiple AI agents (OpenAI, Claude, Gemini, etc.) from a single interface

  • Intelligent Routing: Direct conversations to appropriate agents based on context, rules, and intent

  • Business Integration: Connect AI agents with your existing systems (CRMs, databases, APIs)

  • Conversation Management: Maintain stateful, context-aware conversations across agent interactions

  • Behavior Control: Define complex orchestration logic without coding

Architecture:

  • Lifecycle Pipeline: Configurable processing pipeline (Input → Parse → Route → Agents → Aggregate → Output)

  • Composable Agents: Agents are assembled from reusable, version-controlled configurations

  • Stateful Orchestration: Complete conversation history maintained across agent interactions

  • Asynchronous Processing: Non-blocking architecture handles thousands of concurrent conversations

Key Capabilities:

  • Multi-Agent Orchestration: Coordinate multiple AI agents in a single conversation flow

  • Conditional Agent Invocation: Decide which agents to call based on business rules

  • Agent Response Aggregation: Combine outputs from multiple agents intelligently

  • Seamless API Integration: Connect agents with traditional REST APIs

  • Pattern-Based Input Processing: Route requests based on vocabulary and patterns

  • Dynamic Output Generation: Template-based responses using agent outputs and business data

  • Composable Bot Model: Bots assembled from version-controlled packages and extensions (Bot → Package → Extension)

  • Support for multiple chatbots, including multiple versions of the same bot, running concurrently

  • Support for Major AI API integrations via langchain4j: OpenAI, Hugging Face (text only), Claude, Gemini, Ollama (and more to come)

hashtag
Documentation

Start with these guides to understand EDDI:

Technical specifications:

  • Resource-/REST-oriented architecture

  • Java Quarkus framework

  • JAX-RS

  • Dependency Injection

  • Prometheus integration (Metrics endpoint)

  • Kubernetes integration (Liveness/Readiness endpoint)

  • MongoDB for storing bot configurations and conversation logs

  • OAuth 2.0 (Keycloak) for authentication and user management

  • HTML, CSS, Javascript (Dashboard)

  • React (Basic Chat UI)

NextArchitecture Overview

Last updated

Was this helpful?