Conductor
conductor-oss.org
- Free plan
- Web
- API
- Self-hosted
What Conductor is
Conductor is an open-source platform for building and coordinating AI agents and workflows on a durable execution engine. It stores workflow and task state so work can recover after worker or infrastructure failures, with retries, timeouts, pause, resume and termination controls. Workflow structures can include nested loops, dynamic branching, sub-workflows and thousands of tasks. AI functions include native LLM tasks, MCP tool discovery and calls, human approval steps, and vector workflows for retrieval-augmented generation. Users can bring agents built with LangGraph, CrewAI, Google ADK or the OpenAI SDK. The project lists SDKs for Java, Python, TypeScript, .NET, Go, Ruby and Rust, and documents event-driven orchestration and Agent2Agent connections between independently deployed agents. Conductor is Apache 2.0 licensed and can be self-hosted where Docker runs. Redis is the default storage layer and Elasticsearch handles indexing; storage and queue backends can be replaced. The project also lists a hosted Developer Edition. Both the self-hosted OSS and hosted Developer Edition plans are free.
Who it suits
Conductor is designed for developers who write code and want to build workers and tasks in a preferred language. Teams that need to self-host can run it on infrastructure where Docker runs.
Good to know, and what to watch
- Apache 2.0 licensed and self-hostable.
- Supports retries and recovery after failures.
- Offers SDKs in seven listed languages.
- Includes human approvals and MCP tool calls.
- Designed for developers who write code.
- Redis and Elasticsearch are the defaults for storage and indexing.
Plans
Every plan and priceConductor OSS
Orkes Conductor Developer Edition
Free plan
The verdict
Conductor combines durable workflow execution with agent-oriented features and multiple language SDKs. Its free self-hosted option and hosted Developer Edition offer two ways to get started.
What the maker publishes
- Agent frameworks
- The site says users can bring agents from LangGraph, CrewAI, Google ADK, and the OpenAI SDK. source
- AI capabilities
- Conductor provides native LLM tasks, MCP tool discovery and calls, human approval, and vector workflows for retrieval-augmented generation. source
- Deployment
- The project documents self-hosting with Docker and says Conductor runs anywhere Docker runs. source
- Durable execution
- Conductor persists workflow and task state and supports recovery after worker or infrastructure failure, retries, timeouts, pause, resume, and termination. source
- Integrations
- The integration guide describes event-driven orchestration, MCP tool integration, and Agent2Agent integration with independently deployed agents. source
- Intended users
- The FAQ says Conductor is designed for developers who write code, with workers and tasks built in a preferred programming language. source
- Languages
- The project lists SDKs for Java, Python, TypeScript, .NET, Go, Ruby, and Rust. source
- License
- Conductor is Apache 2.0 licensed and can be self-hosted on your own infrastructure. source
- Storage
- Conductor uses Redis by default and Elasticsearch for indexing, and its storage layers can be replaced with alternative backends and queue providers. source
- Support
- The project directs users to its public Slack community to ask questions and share resources. source
- What it does
- Conductor is an open-source platform for building and orchestrating AI agents and workflows on a durable execution engine. source
- Workflow scale
- The documentation says Conductor supports nested loops, dynamic branching, sub-workflows, and workflows with thousands of tasks. source
Checked against the maker's pages on 7 Oct 2026.