Aegra - Open Source LangGraph Platform Alternative (github.com)

Aegra, a new open-source platform, offers a self-hosted alternative to LangGraph, allowing developers to maintain total control over their AI agent infrastructure without vendor lock-in. Built using FastAPI and PostgreSQL, Aegra complies with the Agent Protocol specification, enabling seamless deployment and orchestration of large language model (LLM) agents. This transition is particularly beneficial for teams prioritizing data sovereignty, customization, and cost management, as they can run Aegra on their own infrastructure at a significantly reduced cost compared to traditional platforms. Notably, Aegra is fully compatible with LangGraph Studio and the existing LangGraph Client SDK, ensuring a smooth transition for current users. It allows for advanced features like human-in-the-loop workflows, full observability through Langfuse, and custom authentication options. Aegra's architecture offers persistent conversation storage and supports interactive debugging, making it a robust solution for developers looking to escape costly vendor ecosystems while maintaining essential functionalities of existing LangGraph applications. With a quick five-minute Docker deployment and comprehensive documentation, Aegra is positioned to enhance development workflows in the AI/ML community.