Orchestra

Orchestra is a managed data orchestration platform for lean data teams, running dbt, Python and AI agent workflows as SaaS or on their own cloud.

Orchestra schedules, runs and monitors the pipelines a data team depends on, from one control plane that gathers metadata from every connected tool. It is built by Orchestra Technologies Limited, an independent company with its registered office in London. Pipelines are written in YAML kept in Git or assembled in a visual builder. Orchestra hosts the control plane, so there is no scheduler to install or patch. The Orchestration module draws each pipeline as a graph of tasks and starts it from cron schedules, sensors that watch for a new file or a query result, webhooks or other pipelines. Tasks retry with exponential backoff, and MetaEngine runs one task group for every item in a list. Lineage shows how tables, views and dashboards depend on each other across the stack. Observability ties each task run to its logs and error messages and sends alerts to Slack, email or PagerDuty. Data Quality runs SQL tests inside the pipeline, so a failed test stops the work that depends on it. The Agent Runtime builds, runs and monitors AI agents, and an MCP interface gives them the metadata Orchestra has collected. Orchestra is built for lean data teams, from a single analytics engineer running dbt to a growing group that has outgrown a self-run scheduler. Teams on dbt with Snowflake or Databricks get the most from its integrations. Larger companies use it for Okta single sign-on, role-based access control and several workspaces. Those that must keep data processing inside their own AWS, Azure or GCP account can run compute there while Orchestra runs the control plane. A typical pipeline loads data with Fivetran or Estuary, transforms it with dbt in Snowflake and then refreshes the dashboards that read from it. Changes move from a development environment to staging and then to production through Git, with GitHub Actions running the staging check on each branch. When a task fails, the alert reaches the on-call engineer in PagerDuty, and the lineage view shows which dashboards are affected. Teams moving off Airflow, Prefect or dbt Cloud use it to bring their scheduling, alerting and lineage into one place.