Skip to main content

​
Workflows API

A workflow is a saved research pipeline. You compose it from a closed catalogue of steps, pin the companies it runs against, publish a version, and then run it on demand, on a schedule, or from an event. Workflows is part of Briefed Research. It is a separate surface from Briefed Intelligence’s signals and markets API and from the Atlas ownership API.

​
Base URL

All routes in this reference sit beneath /workflows.

​
What a workflow is made of

A published workflow version holds a graph: nodes joined by edges. Parts are typed. Each declares what it requires and what it produces, and an edge is only legal when the upstream part produces something the downstream part requires. That contract is enforced identically when you validate and when you save, so a graph that validates will save. See Definitions for the graph shape and Catalogue for the available parts.

​
The lifecycle

Authoring and running are deliberately separate. A definition is inert until a version of it is published, and only a published version can run.
Quickstart walks the whole loop with curl.

​
Access

Workflows requires a Pro, Max or Team Briefed Research plan. A caller without it receives 402 plan_required. Running a workflow draws on your research usage allowance, so API keys carry separate read and write scopes. A key can be issued that reads definitions and run history but cannot publish or run. See Authentication.

​
Response shape

Routes return their payload directly rather than a wrapper. A create returns the object it created:
Errors carry an error code and, where the failure is structured, the specific faults:
See Errors for the full list.

​
Honest degrade

Briefed Research does not fabricate financial data. Where a run cannot obtain evidence, it reports that state rather than substituting a plausible value, and the evidence gate decides whether a below-floor result is held or delivered. A null in a response is meaningful.