Haystack

What it does

Haystack is an open-source Python framework for building LLM applications. It takes in documents, data sources, prompts, and model inputs, then orchestrates them through pipelines and agent workflows that produce answers, retrieved context, generated content, and tool-driven outputs. Developers wire together components for retrieval, routing, memory, and generation, and can run pipelines synchronously or asynchronously with token streaming.

Teams typically use Haystack to build retrieval-augmented generation systems, semantic search, question answering, and autonomous agents. Its distinct approach is explicit control: pipelines are graphs the developer defines and can inspect, while agents let the LLM choose steps, and the two can be nested. Agents expose lifecycle hooks and track token usage and tool calls, and Hayhooks can expose pipelines as REST or MCP endpoints.

Haystack suits developers and engineering teams comfortable with Python who need transparent, customizable LLM systems. It installs with pip install haystack-ai and is Apache-2.0 licensed. Developers supply their own compute and model access. Commercial Enterprise Starter and Enterprise Platform offerings have separate support and hosting terms; building a working application still requires engineering effort.

What it can do

• Build RAG pipelines with hybrid retrieval and self-correction loops
• Create agents with standardized tool calling, lifecycle hooks, and step and token tracking
• Compose branching and looping pipelines with conditional logic
• Run pipelines synchronously or asynchronously with token-by-token streaming
• Process multimodal inputs such as images and audio transcription
• Serialize and deploy pipelines as REST APIs or MCP servers via Hayhooks

Works with

• OpenAI
• Anthropic
• Mistral
• Cohere
• Hugging Face
• Google
• Azure OpenAI
• AWS Bedrock
• Weaviate
• Pinecone

Example workflows

Retrieval-augmented question answering

Inputs: a document collection and a user question. Steps: index documents into a supported store, build a pipeline with retrieval, ranking, and a generator, then run the query. Output: a generated answer grounded in retrieved passages.

Tool-using research agent

Inputs: a research question and a set of tools. Steps: define tools, create an agent or use an Agent Pack template, add lifecycle hooks for guardrails, and run it. Output: a final response plus logged step counts, token usage, and tool calls.

Before you choose

• Requires Python and engineering effort; it is a framework, not a no-code tool.
• Enterprise Starter and Enterprise Platform are separate commercial offerings with their own terms.
• Haystack collects anonymous usage telemetry on pipeline component initialization; it can be opted out of per the docs.

Listed under

• Coding
• AI agents
• Developers
• Enterprise
• LLM framework
• Open-source
• Pipelines

Get started

Follow the quick-start guide after installing the package.

1. Install the package with pip install haystack-ai (or pip install --pre haystack-ai for nightly pre-releases).
2. Work through the Get Started guide to build a first RAG pipeline or tool-calling agent.
3. Connect a model provider and document store, then deploy the pipeline as an HTTP endpoint with Hayhooks if needed.

The software can be installed under its repository license. Infrastructure and model-provider charges remain separate. Optional commercial hosting and support have their own terms.

Haystack is free to use. See their pricing page ↗

Sources for this overview

Product pages, documentation, and pricing checked 29 Oct 2026. Each source supports part of the overview; availability and prices can change.

• Haystack | Open-Source AI Framework for Production-Grade Agents ↗haystack.deepset.ai
• GitHub - deepset-ai/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. · GitHub ↗github.com
• Introduction to Haystack | Haystack Documentation ↗docs.haystack.deepset.ai
• Installation | Haystack Documentation ↗docs.haystack.deepset.ai

Company website

haystack.deepset.ai: product, pricing and policy pages

YC directory mirror

Batch and company name

Web search

Confirms the official site

Social accounts

Accounts the company links, and its posts on X

Human review

Checked by a person

Last checked 29 Oct 2026. Every field keeps its source and date; unknown means we couldn’t confirm it.