Flow
Visual node-graph editor for repeatable AI pipelines
Flow lets you build an AI pipeline by dragging nodes onto a canvas, wiring outputs to inputs and pressing Run. It executes the graph in dependency order, shows every node's result inline, and hands outputs to Studio, SyncCut, DocMan, Dashboard and Kanban. Flows, reusable workflow templates and run records are JSON files in the project workspace, so they sync across devices.
Who it is for: Engineers, researchers and creators who chain several AI, data and publishing steps and want them repeatable and inspectable.
Quick answers
Can I schedule a flow to run every night?
Not from Flow itself; runs execute in the browser tab. Save the flow as a Workflow and create a Kanban task of type Run Flow Workflow, which runs it on the server's orchestrator and links back to the results.
Which AI provider do the nodes use?
ai-generate and generate-script use the backend selected at the top-right of the page (platform, BYOK, local LLM or local CLI via TektBridge). Server-side pipeline nodes use the server's default provider, and write-scenario/write-article assemble text without a model.
Why does Flow have its own project dropdown?
Flow is exempt from the Hub-only picker because flows are project files you often reuse across projects. The dropdown stays in sync with the Hub through a browser channel.
Why are the social post nodes marked queued?
The social posting route is not wired yet; post-* nodes return a queued result with the prepared content. Connector-based publishers need a configured connector.
Where are my flows stored?
In the project workspace: .flow-state.json for the current flow, .flow-template-<name>.json for saved workflows (or _global_flow_templates for global ones), and flow-runs/ for run records. They sync with the project.
Loading the user manual…