GenAI Stack is a Docker Compose configuration that helps developers quickly build generative AI applications by combining Langchain, Neo4j, and Ollama.
The stack addresses the complexity of setting up a complete generative AI development environment by providing a pre-configured Docker composition that integrates a language model backend, a graph database, and an LLM orchestration framework. It supports multiple LLM options including locally-run Ollama models on macOS and Linux, as well as cloud-hosted models like GPT-4, GPT-3.5, and Claude v2 via their respective APIs. The configuration is managed through environment variables, allowing developers to swap components without modifying application code. Embedding models can similarly be chosen from multiple providers including sentence transformers, OpenAI, AWS, Ollama, and Google's GenAI service.
Developers should choose this stack if they want to prototype or build generative AI applications with a graph database backend and need flexibility in model selection. It suits projects that benefit from knowledge graphs and semantic search capabilities. The stack is particularly valuable for those wanting to run models locally via Ollama on macOS and Linux without additional installation steps beyond Docker, though Windows users must install Ollama separately or use cloud-hosted models. The tool includes demo applications that can serve as starting points for custom development.
The project maintains active development with ongoing refinement of its Docker composition and configuration system. The codebase includes watch mode functionality for development workflows, allowing automatic rebuilding when files change. The maintainers have documented a known performance issue in specific Docker Desktop versions and recommend upgrading to resolve it. The stack provides platform-specific profiles for Linux and Linux with GPU support, indicating attention to different deployment environments and hardware configurations.
