Kestra

What Is Kestra?

A data engineer at Leroy Merlin needed to manage increasingly complex workflows and found the existing options either too rigid or too tied to a single programming language. That frustration, in 2019, became the seed of Kestra: an orchestration platform built around declarative YAML instead of Python DAG code. Ludovic Dehon and Emmanuel Darras turned the idea into Kestra Technologies, a company now based near Lille, France.

Kestra sits in the data-orchestration category alongside Apache Airflow, Prefect, and Dagster, but takes a different architectural stance. Where Airflow requires Python to define every pipeline, Kestra treats workflows as YAML documents that can trigger tasks written in Python, Bash, Node.js, Go, or any containerised runtime. The platform is event-driven by default rather than cron-only, and every workflow is versioned and deployable through standard Git-based CI/CD.

Its open-source core is genuinely unrestricted — unlimited flows, unlimited executions, no artificial ceiling pushing you toward a paid tier. On top of that free base, Kestra Technologies sells two paid layers. A usage-based Cloud offering suits teams that don't want to run their own infrastructure. Self-hosted Enterprise Edition adds RBAC, multi-tenancy, and SLA-backed support for larger organisations instead. The company raised a $25M (€21M) Series A in March 2026 led by RTP Global, bringing total funding to roughly €31M ($36M).

Key Features

Declarative, Language-Agnostic Workflows

Every Kestra workflow is a YAML file describing tasks, triggers, retries, and dependencies. That declarative model means a workflow reviewer can read what a pipeline does without parsing Python control flow, and it plays cleanly with Git-based review processes. Because tasks can execute in almost any language or container, a single workflow can call a Python script, a Bash command, and a containerised job without gluing together separate tools.

The Plugin Ecosystem

Kestra ships with more than 1,900 plugins spanning cloud providers (AWS, GCP, Azure), data tools (dbt, Airbyte, Snowflake, Databricks), infrastructure tools (Terraform, Kubernetes), messaging (Slack, PagerDuty), and AI models (OpenAI, Mistral, AWS Bedrock). This breadth reduces the amount of custom connector code teams have to write and maintain, which is often the hidden cost of running an orchestrator at scale.

Event-Driven Triggers

Beyond scheduled cron jobs, Kestra supports webhook, message-queue, and file-based event triggers, so workflows can react to real-time signals rather than waiting for the next scheduled run. This matters for teams orchestrating both batch ETL and near-real-time data or AI pipelines on the same platform, instead of running two separate systems.

AI Copilot and Governance

Kestra's built-in AI Copilot assists with workflow authoring, and the platform includes retries, timeouts, SLAs, audit logs, and role-based access control as core primitives rather than bolted-on extras. Git-driven deployment means a workflow change goes through the same review process as application code, which platform teams tend to value more than a purely visual, drag-and-drop builder.

Hybrid Deployment Model

Because the same engine runs open source, Cloud, and Enterprise Edition, teams can start self-hosted on Docker or Kubernetes and later move to managed Cloud hosting, or vice versa, without rewriting workflows. Air-gapped deployment is also supported for organisations that cannot connect to the public internet at all.

Pricing

Kestra's free tier is the strongest argument in its favour: fully-featured, self-hosted, no execution caps, and the entire 1,900-plugin catalogue included. For a platform team comfortable running its own Docker or Kubernetes cluster, this compares favourably to Prefect Cloud (from roughly $500/month) or managed Airflow deployments in a similar price range for mid-market pipeline volumes.

Kestra Cloud charges on a pay-as-you-go basis for teams that would rather not operate infrastructure themselves, adding automatic scaling, managed upgrades, and SOC 2-backed governance. Neither Cloud nor Enterprise Edition pricing is published on the website; both require a sales conversation, which is a reasonable trade-off for enterprise-grade support but does add friction for teams trying to budget early. Enterprise Edition is licensed per self-hosted instance and layers on LDAP, SCIM, custom RBAC, multi-tenancy, and an SLA-backed customer success program.

EU Compliance & Privacy

Kestra Technologies is registered in La Madeleine, near Lille, placing the company under French and EU jurisdiction. The clearest compliance advantage, though, comes from the architecture itself: self-hosted Kestra keeps all workflow data, credentials, and execution logs on infrastructure the customer controls, satisfying most EU data residency requirements without any special configuration. Air-gapped deployment support extends that further for organisations with strict network isolation rules.

For teams that prefer a managed service, Kestra Cloud is SOC 2 compliant and includes built-in governance and access controls. In the self-hosted model, Kestra doesn't process customer business data itself; it only orchestrates workflows that run on the customer's own infrastructure. That architecture means data sovereignty questions are largely answered by where you choose to deploy it.

Who It's Best For

Data engineering teams frustrated with Airflow's Python-only DAG model get a lower-friction alternative in Kestra's YAML-first, multi-language design, without giving up retries, SLAs, or scheduling maturity.

Platform teams that need both data and infrastructure orchestration benefit from Kestra's plugin breadth across Terraform, Kubernetes, and data tools in one system, instead of running separate orchestrators for each domain.

EU organisations with strict data residency rules should default to self-hosted open source, which keeps everything in-house, over managed Cloud, which processes execution metadata through Kestra's infrastructure.

Teams automating SaaS-to-SaaS business processes — approvals, CRM updates, marketing workflows — are better served by n8n or Make, both EU-built tools designed for that use case, than by Kestra's engineering-grade orchestration model.

The Verdict

Kestra earns a strong recommendation for data and platform engineering teams that want Airflow-class orchestration without Airflow's Python-DAG rigidity, and without paying for a managed service they don't need. The 1,900-plugin ecosystem and genuinely unrestricted open-source tier are real differentiators, not marketing gloss. There are real trade-offs too: community size and battle-tested maturity still favour Airflow, and free-tier support runs through Slack and GitHub rather than a guaranteed SLA. For teams willing to accept a younger ecosystem in exchange for a friendlier declarative model, Kestra is one of the more credible open-source orchestrators built in Europe today.

Frequently Asked Questions

Is Kestra really free to use?

Yes. The open-source core is free under an Apache 2.0 licence with unlimited flows and executions, self-hosted on your own infrastructure. Two paid tiers, Cloud (usage-based) and Enterprise Edition (annual licence), add managed hosting or enterprise governance features on top of the same engine.

Is Kestra GDPR compliant?

Kestra Technologies is a French company near Lille, and self-hosted deployments keep all workflow data on infrastructure you control, which satisfies most EU data residency requirements. The managed Cloud tier is SOC 2 compliant for teams that prefer not to self-host.

How is Kestra different from Apache Airflow?

Declarative YAML replaces Python DAG code, orchestration is event-driven rather than cron-only, and tasks can run in multiple languages out of the box. Airflow has a decade of maturity and a larger plugin community, but Kestra's config-first model is generally faster to onboard for platform teams.

Can Kestra orchestrate AI and infrastructure workflows, not just data pipelines?

Yes. Alongside data orchestration, Kestra supports infrastructure automation (Terraform, Ansible, Kubernetes) and AI workflow automation, including RAG pipelines and agent orchestration through its AI Copilot and plugin catalogue.

How does Kestra compare to n8n or Make for automation?

n8n and Make target business and SaaS-to-SaaS automation with visual, low-code builders. Kestra is built for data-engineering-grade orchestration — versioned YAML, retries, SLAs, and large-scale pipeline execution — making it a better fit for engineering teams than marketing or ops automation.