Reusable workflows for product content.
Build the pipeline once: ingest, AI cleanup, review, metadata, and publishing. Run it by hand, or start it from an upload, a schedule, a webhook, or a product change.
Four steps to a reusable pipeline
Compose nodes once, save the workflow, and run it on the next batch.
Five ways to start a run.
Every trigger runs a saved workflow inside your workspace, with dedupe, an optional credit cap, and a dry run before you turn it on.
Six action families to compose.
Drag nodes onto the canvas and wire them together: AI ops, routing, review, publish, write-back, and API starts.
Six workflow patterns you can build.
Each pattern uses nodes that exist today. Build it once, then run it by hand or attach a trigger.
On demand, on events, or on a schedule.
Run saved pipelines on demand, or attach triggers. Upload, webhook, and product change triggers react to events. Schedule triggers run on a clock.
Production-grade by default.
Workflows fail in messy ways when you build them yourself. The runner records history and retries transient failures. Triggers add dedupe, credit caps, and loop protection.
Common questions.
Related reading
Where to go next, depending on which side of the workflow you're building first.