Hatchet is an open-source distributed task queue for running background work and durable multi-step workflows on infrastructure you control or on its managed cloud. It serves backend teams that have outgrown a simple job library and need retries, scheduling, concurrency control, and observability without adopting a separate broker and orchestration stack. If your product sends notifications, processes files, bills customers, or drives AI agents through long-running steps, Hatchet is the engine that executes that work reliably.[1][2]
Quick verdict
Backend teams running background jobs or AI-agent workloads on Postgres-friendly infrastructure will get the most from Hatchet. Its durable execution model with retries, DAG workflows, and autoscaling workers covers the standard reliability workload, and the MIT core self-hosts with Docker or a single-binary Lite mode. The material constraint is its code-first, developer-operated nature, so teams without backend capacity should choose a visual automation platform instead. If you have developers and can self-host or budget usage-based cloud spend, choose Hatchet; if you need visual no-code building, choose n8n.[1][2]
Key features
Postgres-native queues Tasks, events, and schedules persist in Postgres with fair queueing, task affinity, and rate limiting, so there is no separate broker to deploy, secure, or scale.[1]
Durable execution Long-running workflows survive restarts with sleep and resume semantics, retries with backoff, timeouts, and replay support for recovering failed or interrupted runs.[1]
Workflows and DAGs Multi-step workflows compose tasks as directed graphs with child and spawned runs, scheduled and cron triggers, and event-driven starts for pipeline-style work.[1]
Autoscaling workers Workers V2 scale with demand under concurrency and rate-limit controls, with sticky workers for affinity workloads and slot-aware scheduling.[1]
Exactly-once-minded controls Run idempotency, deduplication, and conflict-aware scheduling guard against double execution when producers retry or events redeliver.[1]
Four first-class SDKs Python, TypeScript, Go, and Ruby SDKs define, enqueue, and execute work natively, so each service uses its own language rather than a sidecar DSL.[1]
Operational observability The dashboard surfaces runs, queues, and worker state, with OpenTelemetry tracing and Prometheus plus Grafana metrics for production monitoring.[1]
Self-host or managed cloud The engine self-hosts with Docker, a single-binary Lite mode, or Kubernetes guides, while Hatchet Cloud adds autoscaling, multi-region execution, and SSO on usage-based pricing.[1][2]
Who will love it
Queues, retries, scheduling, and observability arrive as one Postgres-backed system, replacing a patchwork of job libraries and cron hosts.[1]
Durable execution with sleep, resume, and replay fits long-running agent steps, and event triggers plus child runs map onto agent orchestration patterns.[1]
The MIT core self-hosts licence-free on infrastructure they already run, and the cloud free tier covers one million task runs and one million events monthly without a card.[1][2]
Who should avoid it
Workflows are defined in code through SDKs, so teams without developers cannot build or change automations the way a visual platform allows.[1]
Even the Lite binary is something to run and watch, and production self-hosting means operating Postgres plus the engine with real on-call responsibility.[1]
Cloud spend meters per million task runs with overage past the free allowance, so spiky or enormous volumes need the unit math done before committing.[1]
Alternatives and compatibility
n8n wins for visual building with code escape hatches across hundreds of integrations, while Hatchet wins for code-first durable execution with engine-level reliability controls.[1]
Hatchet connects through SDKs and events in your own services, so a team whose work is mostly third-party SaaS wiring will move faster visually.[1]
Stay with Hatchet when background jobs, durable workflows, and worker-level controls are the product; switch toward n8n when operators must assemble and adjust workflows themselves.[1][2]
Questions worth asking
What is the best Hatchet alternative for non-technical users?
n8n is the safer pick when operators must build visually: its canvas, ready-made integrations, and code nodes cover automation without backend development. Hatchet itself is code-first, so non-technical teams should stay with a visual platform rather than adopt SDK-defined workflows.[1]
Is Hatchet really no-code?
No. Workflows, tasks, and workers are defined in code with the Python, TypeScript, Go, or Ruby SDKs, and operating it needs backend and infrastructure skills. The dashboard is a visual surface for monitoring and managing runs, not a builder for creating automations.[1]
How much does Hatchet cost per month?
Hatchet Cloud pricing checked October 11, 2026: the free tier includes one million task runs and one million events per month with no card required. Paid tiers meter task runs per million (observed list figures include thirty dollars per million with volume discounts), include fifty users before a ten-dollar per-user charge, and extend log retention from one day to seven days, with custom plans for longer retention and enterprise terms. Self-hosting the MIT core is licence-free; hosting, Postgres, and operations still cost whatever your infrastructure charges.[1][2]
Can you self-host Hatchet?
Yes. The MIT-licensed engine self-hosts with Docker, a single-binary Lite mode that needs no separate API or database, and Kubernetes deployment guides, with the full engine requiring only Postgres. Hatchet Cloud remains the alternative for teams that prefer managed autoscaling, multi-region execution, and SSO without operating the engine. No installation guide is included here.[1][2]
Sources and editorial review
Source research. Reviewed on October 11, 2026. Written by Sanjib Sah. AI assisted; reviewed by a human editor.
We have not run or tested this product.
- hatchet-dev/hatchet on GitHub
official · Observed October 11, 2026
Identity, capabilities, self-hosting, cloud offering, SDKs, release recency.
- Hatchet Cloud pricing
official · Observed October 11, 2026
Cloud tiers, free allowance, overage rate, users, retention, SSO, regions, support.
Compared with these products
No explicit proprietary baseline is recorded.