The Open Source AI Middleware. EDDI offers a centralized and secure way to manage generative AI chatbots and (conversational) APIs, with the added benefit of keeping track of all AI-related conversations in one place.
E.D.D.I key features
**EDDI Manager**: A visual agent management UI for building, deploying, and monitoring AI agents.
**MCP Server**: Provides 42 tools for AI-native control, enabling interaction with external tools and managing agents programmatically.
**Config-as-Code**: Utilizes JSON configurations for agent logic and pipelines, allowing instant iteration without redeployment.
**Security-First Architecture**: Incorporates features like Vault integration, cryptographic audit trails, and no dynamic code execution to ensure security.
**Performance**: Supports millions of concurrent threads for I/O-bound workloads, optimized for cloud-native environments.
E.D.D.I use cases
**Visual Agent Management**: Utilize the EDDI Manager for creating, configuring, and monitoring AI agents through an intuitive drag-and-drop interface.
**Configuration-as-Code**: Implement agent logic and workflows using JSON configurations, allowing for instant updates without redeployment.
**Multi-Agent Orchestration**: Automatically route user requests to the most suitable agent based on intent, enabling managed conversations and A/B testing.
**Observability and Audit Trails**: Track every pipeline step with immutable logs, including token usage and performance metrics, ensuring compliance and operational monitoring.
**Security-First Architecture**: Leverage built-in security features such as OIDC/Keycloak authentication and cryptographic audit trails to protect against vulnerabilities.
Useful for
**Visual Agent Management**: EDDI Manager provides an intuitive UI for building, deploying, and monitoring AI agents, enabling non-developers to manage agent behavior without coding.
**Configuration-as-Code**: Allows for instant iteration on agent logic and pipelines using JSON configurations, eliminating the need for redeployment and making it easier for prompt engineers to modify rules.
**Security-First Architecture**: Built with a focus on security, EDDI avoids dynamic code execution, integrates with secure vaults, and maintains cryptographic audit trails, ensuring compliance and reducing vulnerabilities.
**High Performance**: Supports millions of concurrent threads for I/O-bound workloads, ensuring efficient processing and responsiveness in enterprise environments.
**Comprehensive Observability**: Provides immutable audit trails and detailed logging of every pipeline step, enabling effective monitoring, operational insights, and regulatory compliance.
Price
EDDI is a self-hosted enterprise AI platform available as version 6.
It features a configuration-driven approach using JSON for agent logic, pipelines, and tool definitions.
The platform includes 42 tools for AI-native control via the Model Context Protocol (MCP).
EDDI is built for performance with millions of concurrent threads and supports both MongoDB and PostgreSQL.
It is open-source, licensed under Apache 2.0, and has passed over 5,100 tests with zero failures.