HTTP 200 verified twice
Finding your next stop…
Loading the latest directory information.
Loading the latest directory information.
Start here for the decision-making essentials: what Haystack does, who it is for, how it is accessed, and the first-party sources behind this profile.
Enterprise Support offers flexible pricing based on company size
Create agentic, context engineered AI systems using Haystack’s modular and customizable building blocks, built for real-world, production-ready applications.
Read-only public captures of Haystack’s homepage. Screenshots are dated, never live embeds, and open full-screen.
Short answers to the questions buyers and builders commonly ask about Haystack. Each answer cites the shared ledger below, where every source is listed once.
Modular framework that provides full visibility to inspect, debug, and optimize every AI d · Orchestrates every step of an AI agent, from retrieval to reasoning to memory and tool use · Built-in reliability and observability to run production workloads across any environment · CompactionHook provides experimental context compaction for Agent that runs before LLM cal
Haystack's modular framework gives you full visibility to inspect, debug, and optimize every decision your AI makes.
Build real-world, production-ready agentic and context-engineered AI applications
built for real-world, production-ready applications
PipelineSnapshot.pipeline_state.inputs shape changed from {component: {socket: value}} to
PipelineSnapshot.pipeline_state.inputschanged shape. It used to store one flattened value per socket,{component: {socket: value}}. It now stores the pipeline's internal inputs, keeping the component that sent each one in the order it arrived
Enterprise Support offers flexible pricing based on company size
Flexible pricing based on company size
184 total integrations available · Connects to OpenAI, Anthropic, Mistral, Hugging Face, Weaviate, Pinecone, Elasticsearch an · OpenAI integration via OpenAITokenCounter calling OpenAI's counting API for exact, model-s
184 Total Integrations
Pipelines are serializable, cloud-agnostic, and Kubernetes-ready with logging, monitoring,
Haystack Pipelines are serializable, cloud-agnostic, and Kubernetes-ready, with logging, monitoring, and deployment guides to support you.
This profile connects the jobs Haystack is described as handling with its delivery model, access options and the subjects used to match it to related products in this directory.
Each fact points to a recorded source, making it easy to distinguish verified product information from claims that need checking.
HTTP 200 verified twice
Haystack | Haystack
Modular framework with full visibility to inspect, debug, and optimize AI decisions · Orchestrates retrieval, reasoning, memory, and tool use for AI agents · Built-in reliability and observability for production workloads in any environment
Modular framework that provides full visibility to inspect, debug, and optimize every AI decision
Orchestrates every step of an AI agent, from retrieval to reasoning to memory and tool use
Built-in reliability and observability to run production workloads across any environment
Build real-world, production-ready agentic and context-engineered AI applications
AI Orchestration Framework
Pipelines are serializable, cloud-agnostic, and Kubernetes-ready, with logging, monitoring, and deployment guides
184 total integrations available
Connects to OpenAI, Anthropic, Mistral, Hugging Face, Weaviate, Pinecone, Elasticsearch, and more with no vendor lock-in
Distributed from open source to enterprise tiers covering every step of the AI journey
A concise view of the jobs, capabilities and integrations described in the recorded product sources.
Documented product formats, platforms and official distribution destinations. Availability can vary by region and plan.
A concise history of software releases and material product changes. Events appear only when a dated source supports what changed.
Showing the newest updates and meaningful milestones. Open an entry for its summary and source.
Major 3.0.0 release of the open-source Haystack AI orchestration framework was published.
View source [3]Haystack 2.31.0 update was published as part of the active 2.x series.
View source [3]Patch release in the 2.30 line; specific change details are not provided in the supplied index entry.
View source [3]Patch release in the 2.30 line; specific change details are not provided in the supplied index entry.
View source [3]Minor release in the 2.x line; specific change details are not provided in the supplied index entry.
View source [3]Minor release in the 2.x line; specific change details are not provided in the supplied index entry.
View source [3]Minor release in the 2.x line; specific change details are not provided in the supplied index entry.
View source [3]Minor release in the 2.x line; specific change details are not provided in the supplied index entry.
View source [3]Facts, answers, structured details, milestones and primary resource links cite this shared ledger. Each external page appears once; release tags from the same GitHub project are grouped under one release history.
These tools share a workflow, capability or audience with Haystack. They may complement it rather than replace it, and are not presented as integrations or endorsements.
Enterprise Support offers flexible pricing based on company size
Enterprise Support includes private, secure engineering support plus best practices templates and deployment guides
Latest release Haystack 3.1.1 on September 3, 2026
CompactionHook provides experimental context compaction for Agent that runs before LLM calls and shortens the conversation once it crosses a configured fraction of the context window
SlidingWindowCompactor removes entire historical turns first, then individual steps of the current task, replacing removed content with a short omission note
ToolResultPruningCompactor replaces older/large tool results with placeholders while keeping the most recent tool-calling steps (and their parallel results) intact
New haystack.token_counters module with ApproximateTokenCounter (no dependencies), TiktokenCounter (local, closer for OpenAI models), and OpenAITokenCounter (calls OpenAI's counting API for exact, model-specific counts)
AgentTool wraps a Haystack Agent as a Tool so another Agent can delegate to it; only the wrapped Agent's final reply is visible to the caller
Agent.clone() method returns a new Agent with the same configuration, optionally replacing some init parameters
Agent now returns an exit_reason output ('text', the name of the tool that satisfied an exit condition, or 'max_agent_steps') to enable downstream routing, also exposed to hooks via state.get("exit_reason")
PyPDFToDocument and PDFMinerToDocument components support a link_format parameter that parses PDF annotations and appends links at the bottom of page content, matching DOCXToDocument
Document-store components provide close and close_async methods for releasing resources on AutoMergingRetriever, CacheChecker, DocumentWriter, FilterRetriever, and SentenceWindowRetriever
HAYSTACK_UNSAFE_DESERIALIZATION environment variable, when set to a truthy value, process-wide disables all deserialization safety checks (module allowlist, builtin/import-primitive and control-plane denylists, object-internals traversal gu
PipelineSnapshot.pipeline_state.inputs shape changed from {component: {socket: value}} to {component: {socket: [{"sender": ..., "value": ...}]}} with a new PipelineState.inputs_format field; snapshots from earlier versions keep the flattene
OpenAI integration via OpenAITokenCounter calling OpenAI's counting API for exact, model-specific token counts
AI orchestration framework by deepset · Modular and customizable building blocks for production-ready applications · Orchestrates retrieval, reasoning, memory, and tool use
Create agentic, context engineered AI systems using modular and customizable building blocks, built for real-world, production-ready applications
Orchestrate every step of an AI agent, from retrieval to reasoning to memory and tool use
Modular framework with full visibility to inspect, debug, and optimize every decision the AI makes
184 total integrations
Open architecture that lets users mix and match components to fit their workflow
Enterprise support offering includes private secure engineering support, best practices templates, deployment guides, and access to flexible services
Enterprise support is offered with flexible pricing based on company size
Haystack Enterprise Starter: best practices and support program for building and scaling production-grade AI apps
Haystack 3.1.1 released on September 3, 2026 (most recent release shown)
3.1.0
deepset
Context length management for Agents via CompactionHook with SlidingWindowCompactor and ToolResultPruningCompactor strategies
