Workflow Orchestration & Durable Execution brand directory

Temporal is an open-source durable execution platform that lets developers write reliable, fault-tolerant distributed workflows using standard programming languages. It automatically persists all workflow state, retries failed activities, and resumes execution exactly where it left off after any failure—eliminating the need for custom retry logic, state checkpointing, or message queues. Available as a managed cloud service (Temporal Cloud) or self-hosted (MIT license), it is used for AI agent orchestration, payment processing, order management, infrastructure automation, and long-running business processes across thousands of enterprises.

AWS Step Functions is Amazon Web Services' serverless workflow orchestration and durable execution service, enabling teams to build, visualize, and run distributed multi-step applications as state machines integrated across the AWS ecosystem.

Inngest is a managed durable execution and workflow orchestration platform that lets developers make any code fault-tolerant by wrapping it in SDK steps. It replaces queues, schedulers, and state management with a single reliability layer accessed through TypeScript, Python, Go, or Kotlin SDKs. Core primitives include step.run (automatic retry and exactly-once execution per step), step.sleep (delays spanning minutes to months), step.waitForEvent (human-in-the-loop or event-driven pause), and flow-control features such as per-user concurrency keys, throttling, debouncing, and prioritization. The platform ships with a local dev server for instant trace-level debugging, bulk Replay for incident recovery, and AI/agent tooling via AgentKit and MCP integration.

Prefect provides a Python-native workflow orchestration framework (open-source, Apache 2.0) and two commercial managed platforms: Prefect Cloud for production workflow orchestration with enterprise security, autoscaling workers, and observability; and Prefect Horizon, a managed AI infrastructure platform for deploying MCP servers, managing agent context delivery, and governing AI integrations. The open-source core requires Python 3.10+ and supports self-hosted deployments via Docker, Kubernetes, and Helm. A patented hybrid execution model separates orchestration from execution, allowing user code and data to remain on-premises while the control plane is cloud-hosted.

Orkes is an enterprise agentic workflow orchestration platform built on the Netflix Conductor open-source engine. It enables teams to design, execute, and monitor complex workflows across microservices, AI agents, APIs, and human tasks—with built-in durability, observability, and enterprise-grade governance—delivered as a fully managed cloud service or customer-hosted deployment on any major cloud or on-premises.

Restate is a durable execution platform that transforms ordinary application code into fault-tolerant, automatically-resuming processes. Developers wrap side effects in ctx.run() calls; Restate journals each completed step to its embedded distributed log, enabling automatic replay from the last checkpoint after any infrastructure failure—eliminating the need for manual retry logic, saga boilerplate, or distributed locks. Beyond workflows, Restate introduces Virtual Objects (actor-style stateful entities with isolated K/V storage) and durable exactly-once RPC, making it a general-purpose resilient distributed programming model. The runtime is written in Rust, ships as a single binary with no external dependencies, and achieves sub-100ms p99 latency for multi-step workflows. It supports six language SDKs and runs across serverless FaaS platforms, containers, and Kubernetes. Restate Cloud provides a managed, SOC-2-certified version with an observability UI, cloud tunneling for local development, and tiered pricing based on durable action consumption.

Trigger.dev is a fully-managed, open-source platform for writing and deploying durable AI agents and background workflows in TypeScript. Developers define tasks as ordinary async functions inside a /trigger folder in their codebase; the SDK, CLI, and cloud handle packaging, deployment, queuing, retries, scheduling, checkpointing, and observability. Key differentiators include true no-timeout task execution via a checkpoint-resume system, built-in AI agent patterns (tool calling, parallelisation, human-in-the-loop), a Realtime API for streaming task progress and LLM output to frontends, and an extensive build-extension system (FFmpeg, Prisma, Python, Playwright, system packages). Available as a managed cloud (Trigger.dev Cloud) or self-hosted via Docker/Kubernetes under Apache 2.0.

Windmill is an open-source, self-hostable developer platform and workflow engine that transforms scripts in mainstream languages into production-grade internal tools, multi-step DAG workflows, scheduled jobs, and custom UIs—without proprietary SDKs, vendor lock-in, or infrastructure boilerplate.

Kestra is a declarative, event-driven, open-source orchestration platform that unifies data pipeline, infrastructure automation, AI workflow, and business process orchestration under a single YAML-based control plane, supported by 1,300+ plugins, a visual UI with real-time DAG visualization, and enterprise governance capabilities including RBAC, multi-tenancy, audit logs, and air-gapped deployment.

Camunda is an enterprise-grade workflow orchestration and durable execution platform built on open BPMN and DMN standards, combining visual process design tools for business and IT stakeholders with a high-throughput, cloud-native Zeebe engine for executing complex, long-running processes at scale—including AI agent coordination, microservices choreography, human task workflows, and legacy system modernization.

Hatchet is a durable workflow orchestration engine and background task platform built on Postgres. It enables developers to define tasks and workflows as code, execute them with fault tolerance and automatic retries, and observe them through a real-time dashboard. It is available as a managed cloud service or a self-hosted open-source deployment.