Orloj
An orchestration runtime for multi-agent AI systems. Declare agents, tools, and policies as YAML; Orloj schedules, executes, routes, and governs them for production-grade operation.
Last verified:
What is Orloj?
Orloj is an open-source orchestration runtime for multi-agent AI systems that treats agent infrastructure as code. Define agents, tools, policies, and workflows in declarative YAML manifests, then let Orloj handle scheduling, execution, model routing, governance enforcement, and reliability so you can run multi-agent systems in production with operational rigor comparable to Kubernetes for containers.
The platform uses a server/worker architecture where orlojd (server) hosts the REST API, resource store (in-memory or Postgres), and task scheduler, while orlojworker instances claim and execute tasks, route model requests through gateways (OpenAI, Anthropic, Azure OpenAI, Ollama), and run tools in configurable isolation (sandboxed, container, or WASM). Governance is enforced inline at the execution layer through AgentPolicy, AgentRole, and ToolPermission resources that fail closed on unauthorized actions.
Key capabilities include DAG-based orchestration with pipeline/hierarchical/swarm-loop topologies, per-agent model binding, tool isolation with timeouts and retry, cron-based scheduling and webhook-triggered task creation, lease-based task ownership with capped exponential retry and jitter, idempotency tracking, dead-letter handling, full task trace and message lifecycle logging, per-agent/per-edge metrics, and a built-in web console with topology views and task inspection.
Orloj is designed for engineering teams building production multi-agent systems who need deterministic execution, policy enforcement, and safe operations under real production load. It's particularly valuable for organizations moving past single-agent demos into production agent fleets with governance requirements for compliance frameworks like HIPAA, SOC 2, and the EU AI Act. Infra engineers will appreciate the YAML-first workflow, CLI apply/rollback, and version-controlled manifests.
Orloj pricing
Pricing model: Freemium
Orloj is free and open source under Apache 2.0 with the full runtime available forever. The Community tier includes unlimited agents and workflows, DAG orchestration, fan-out/fan-in, fail-closed governance engine, MCP tool integration, built-in observability console, CLI and YAML-first workflow, and community support via Discord. The Cloud tier is pay-as-you go with usage-based pricing featuring fully managed control plane, on-demand worker scaling, hosted dashboard with logs and traces, automatic runtime updates, team workspaces, and email/chat support (coming soon). The Enterprise tier is custom annual contract including everything in Cloud plus SSO/SAML authentication, RBAC, audit log export and SIEM integration, custom SLAs and uptime guarantees, dedicated support engineer, and on-prem/VPC deployment option.
Orloj pros
- Open source under Apache 2.0 license forever
- Declarative YAML manifests for agents, tools, policies, and workflows
- Kubernetes-like operational rigor for AI agents
- Built-in fail-closed governance enforced at execution layer
- DAG-based orchestration with fan-out/fan-in support
- Lease-based task ownership prevents conflicts
- Capped exponential retry with jitter for reliability
- Dead-letter queue handling for failed tasks
- Idempotency tracking for safe replay
- Per-agent token caps and model allowlists for cost control
- Tool isolation via container, WASM, or sandboxed execution
- Built-in web console with topology views and task inspection
- Full task trace and message lifecycle logging for auditability
- Model routing support for OpenAI, Anthropic, Azure OpenAI, Ollama
- CLI with orlojctl for apply/rollback workflow
- Cron-based scheduling and webhook-triggered task creation
- Works with agents from any framework via standardized tool interfaces
- Single-process mode with embedded worker for local development
- Postgres storage backend for production deployments
- Per-agent/per-edge metrics and live event streaming
Orloj cons
- Early-stage project still evolving quickly
- Requires learning YAML manifest syntax and new concepts
- Built-in primarily for Go developers contributing code
- Managed cloud version still coming soon (not available yet)
- Full production stack requires Postgres and optionally NATS
- Some templates like incident response still marked coming soon
- Steep learning curve for teams unfamiliar with infrastructure-as-code
- No native GUI for editing manifests (YAML-first approach)
- Community support via Discord only (no enterprise support for free tier)
- Limited documentation compared to mature orchestration tools
- Tool approval process still pending for some features
- Requires understanding of agent policies, roles, and permissions
- Self-hosted option requires managing your own infrastructure
- Agent evaluation features may need setup for golden datasets
- Not suitable for simple single-agent use cases
Frequently asked questions about Orloj
What is Orloj agent orchestration?
Orloj agent orchestration is coordinating multiple AI agents in production with governance, scheduling, and observability. It's like Kubernetes for agents: you need the same operational rigor as you do for containers or databases. Orloj provides an orchestration plane purpose-built for AI agent systems where agents become manageable infrastructure declared in version-controlled YAML manifests.
How does Orloj compare to LangChain and CrewAI?
LangChain helps you build agents. CrewAI helps agents collaborate. Orloj runs agents in production, with governance, observability, and the reliability patterns you expect from infrastructure. They're all solutions to different problems, not competing. Orloj is an orchestration plane for running agents you've built with any framework.
What does fail-closed governance mean?
Fail-closed means unauthorized actions are denied by default. An agent can only use tools you explicitly permit. Fail-open (the alternative) would allow actions unless you explicitly block them, which is a risky default in production. Orloj enforces policies and permissions inline on every turn and tool call, with unauthorized actions failing closed with traceable outcomes.
Can I use Orloj with agents built in other frameworks?
Orloj is an orchestration plane for running agents. You can build agents in Orloj just like you would with frameworks like LangChain, LlamaIndex, or CrewAI. Orloj then manages them at scale with governance, scheduling, and reliability. It works with agents built in any framework via standardized tool interfaces, though some refactoring may be needed for specific governance requirements.
Do I need to rebuild my agents from scratch to use Orloj?
Not necessarily. Orloj works with agents built in any framework via standardized tool interfaces. Some refactoring may be needed for specific governance requirements, but you don't need to rebuild from scratch. You can refactor existing agents to work with Orloj's declarative YAML manifests and governance layer.
How does Orloj handle agent failures and retries?
Orloj includes lease-based task ownership, retry with jitter, idempotency tracking, and dead-letter handling. These patterns prevent cascading failures and ensure your agent fleet survives partial outages. The runtime handles reliability primitives you'd otherwise hand-roll, with bounded retry and dead-letter transitions built into the core.
Is Orloj compliant with regulations like HIPAA and SOC 2?
Orloj logs all agent actions, tool calls, and policy decisions. The structured audit trail is designed to support compliance workflows for frameworks like HIPAA, SOC 2, and the EU AI Act. Governance is enforced at the execution layer, not as an afterthought, with full task trace and message lifecycle logging.
What observability features does Orloj provide?
Orloj provides structured logging, distributed tracing, metrics collection, and cost attribution. You can trace an agent's decision path, see which tools it called, understand latency, and allocate costs by agent or workflow. The built-in web console offers topology views, task inspection, live event streaming, and per-agent/per-edge metrics.
Can I run Orloj on-premise or in my own VPC?
Yes. Orloj is Apache 2.0 licensed and developed publicly on GitHub. You can run it on-premise or in your own VPC. The full runtime is open source including governance, orchestration, scheduling, and observability, with no limits on agents or workflows when self-hosted.
Is Orloj difficult to learn for engineers?
If you're familiar with Kubernetes, Docker, or infrastructure-as-code tools, Orloj will feel familiar. You define agents and policies in YAML manifests and deploy with a single command. The concepts are straightforward for engineers. The CLI uses orlojctl apply -f for declarative deployment similar to kubectl apply.