Runs
Run one node, one branch, or the whole pipeline.
Pick the scope. Each run executes in an isolated sandbox, and its status, logs, schema, and artifacts stay with the nodes it touched.
Every node execution is on the record
Observable by default
Every node execution is on the record
Open any run to see each node's status and duration, then the rows, columns, and transformation steps it produced, next to the script that ran.
- Every node execution in a run, with its duration
- Rows in and rows out for every step
- Manual, agent, and scheduled runs in one history
Runs in detail
01
Three scopes
Run one node, everything downstream of it, or the full pipeline.
02
Isolated sandboxes
Each node executes in its own sandbox, so one failure does not pretend the rest succeeded.
03
Paused for review
A run stops at a node whose new script is waiting on a person, then resumes after approval.
04
Run history
Every run in a project, manual, agent-driven, or scheduled, in one list you can filter.
05
Row counts per step
Transforms log every step with rows in and rows out, like 1,957 to 240 when orders roll up to customers.
06
Keeps going without you
A run is a job on the server. Close the tab and it still finishes, and the canvas catches up when you return.
Related capabilities
Approvals
Inspect before anything is trusted
Node changes and new scripts wait for a person to inspect and approve them.
Explore approvalsCanvas
The pipeline as one editable graph
Sources, joins, transforms, validation, and outputs on one graph you can see and change.
Explore canvasTransformations
Generated Python and SQL on each node
Describe the step. The agent writes Python and SQL from the upstream schema and attaches it to the node.
Explore transformations
Ready to build a pipeline you can inspect?
Describe the outcome, inspect what the agent proposes, and run it when you trust it.