haystack
Open-source AI orchestration framework for building context-engineered, production-ready LLM applications. Design modular pipelines and agent workflows with exp
What is haystack?
Open-source AI orchestration framework for building context-engineered, production-ready LLM applications. Design modular pipelines and agent workflows with explicit control over retrieval, routing, memory, and generation. Built for scalable agents, RAG, multimodal applications, semantic search, and conversational systems.
Overview of haystack
What it is
Haystack is an open-source AI orchestration framework for building production-ready LLM applications in Python. It allows developers to design modular pipelines and agent workflows with explicit control over retrieval, routing, memory, and generation.
Key features
Agents built for production: lifecycle hooks (before_llm, before_tool, on_exit), built-in tracking of step_count, token_usage and tool calls for monitoring and cost control. Ready-made agents from Agent Pack (deep research, advanced RAG) and progressive skill discovery via SkillToolset.Built for context engineering: explicit control over how information is retrieved, ranked, filtered, combined, structured, and routed before reaching the model. Pipelines and agent workflows are transparent and traceable.Native Async Support: a single Pipeline can run synchronously or asynchronously and stream tokens; Agent can run concurrent tool calls.Modular and customizable: built-in components for retrieval, indexing, tool calling, memory, and evaluation; users can create their own components, add loops, branches, and conditional logic.Model- and vendor-agnostic: integrates with OpenAI, Mistral, Anthropic, Cohere, Hugging Face, Google, Azure OpenAI, AWS Bedrock, local models, and others without rewriting the system.Extensible ecosystem: consistent interface for building and sharing custom components, enabling community and third-party extensions.
Use cases
Building scalable RAG systems, multimodal applications, semantic search, question answering, autonomous agents, and conversational systems. Haystack is designed for contexts where retrieval, routing, memory, and generation need to be explicitly controlled and traced.
Getting started
pip install haystack-aiTo install nightly pre-releases:pip install --pre haystack-aiHaystack also supports Docker images. See the repository README for a comprehensive installation guide.
License and pricing
Licensed under Apache-2.0. Haystack Enterprise Starter and Haystack Enterprise Platform are available for expert support, templates, deployment guides, and a managed cloud or self-hosted production setup, with pricing details not specified in the README.
Based on the project's public README and repository metadata.
