Hub Docs User documentation
Back to home
Hub Docs

Processes

What it is

A Process is a visual diagram you build to automate work in your Space. You drag triggers, AI tools, and flow-control nodes onto a canvas, connect them with arrows, and Hub runs the diagram every time the trigger fires.

A process is one of two types:

  • Pipeline — a step-by-step automation that reacts to events, schedules, webhooks, actions, or a click in the editor
  • Product — the rules and steps behind a financial product offered to members. Product processes live inside a Blueprint and are reached through the Blueprints page rather than the Processes list.

When to use it

Reach for Processes when you want to automate work without writing code. Examples:

  • Run a daily summary of open tasks and email it to the team
  • When a new company is created, automatically set up its onboarding tasks
  • Accept data from an external system through a webhook and create Hub records
  • Approve, validate, or post a transaction when a member triggers an action
  • Kick off a one-off run on demand from the editor toolbar

Where to find it

Open the Tenet admin Space and click Processes in the sidebar. The Processes page has its own sidebar with two entries:

  • Pipelines — every Pipeline you've built. Each row shows how many runs the pipeline has accumulated.
  • Runs — every run across every pipeline, with the workflow, trigger, type, status, and start time. Click a row to jump to the live run view.

Product processes are created as part of a Blueprint and don't appear here.

From the Processes page you can:

  • Use the New pipeline button to start a new process
  • Use the Import button to upload a process YAML file; on success Hub takes you straight to the new pipeline's editor
  • Click any pipeline row to open the visual editor
  • Use the Edit action on a row to jump straight into the editor, or the Delete action to remove a pipeline you no longer need (Hub asks you to confirm before it disappears). Delete is only shown to operators with permission to remove pipelines.

Most members of a Space won't see this page — it sits in the operator admin area.

How it works

The canvas. Every process is a graph: each box is a node, each line is the order in which they run. The toolbar at the top of the canvas lets you publish, export, and view issues. The process auto-saves as you work — about a second and a half after your last change, the toolbar shows a "Saved just now" hint. There is no manual Save button.

Each node shows an icon and a coloured stripe that match its category — triggers, AI tools, flow control, side effects — so you can read a process at a glance. The technical type appears in small text below the title.

You can rename a node from the top of its settings panel; the new name appears on the canvas card immediately.

Adding a node. Press / anywhere on the canvas (or click Add node) to open a search-first picker. Start typing what you want to happen — "wait for payment", "send email", "create task" — and Hub ranks matching nodes across all categories. Use the arrow keys and Enter to insert; press Esc to dismiss. The new node lands on the canvas where you can drag it into place.

Editing a node. Click any node to open its settings in a side panel; the canvas shrinks to make room rather than disappear behind a modal. Click another node to hot-swap the panel without closing it. The panel header has a width toggle so you can switch between a narrow and a wide view (Hub remembers your choice).

Connecting nodes. Drag from the right edge of one node to the left edge of the next. Branching nodes have multiple coloured outputs:

  • Conditional — green "then" and red "else"
  • Try / Catch — green "try" and red "catch"
  • Switch — one output per case plus a grey "default"
  • Foreach — a "body" output that runs once per item in the list

Pausing a run. Drop a Wait for events node into the graph to pause the process until something happens. In its settings panel, list the specific events you're waiting for — each must include an ID, e.g. payment.completed.123 — pick All (wait for every one) or Any (continue on the first), set an optional Timeout in hours, and give the wait a short Message describing why it's paused (this shows up on the run while it's parked).

Triggers. Every process starts with a trigger. The trigger types are:

  • Event — fires when something happens in Hub (a company is created, a task is closed, a member is added)
  • Cron — runs on a recurring schedule, like "every Monday at 9am"
  • Webhook — fires when an external system sends an HTTP request to a unique URL Hub generates for the trigger
  • Action — fires when a member runs a specific action like a salary sacrifice, fund switch, or fund withdrawal
  • Payroll — fires on payroll dates for a member
  • Manual — fires on demand when you click Run in the editor toolbar

Run button. When the process has a Manual trigger, the editor toolbar shows a Run button. With one Manual trigger it runs that trigger straight away. With several, Run opens a menu listing each by name so you can pick which one to fire. With none, the button is hidden.

Webhook keys. When you save a process with a webhook trigger, Hub generates a one-time API key. Copy it immediately — Hub only shows it once. To get a new key, click Regenerate on the trigger.

Available variables. When you click into a node, the inspector lists every variable you can reference at that point in the graph. Click a variable chip to copy it to the clipboard, then paste it into a tool input.

Inputs that fit the field. Inputs in the node settings panel adapt to the kind of value they take:

  • Space and User fields use a searchable picker so you can find someone by name. When a node has both a Space and a User picker side by side, the User list filters to people who actually belong to the Space you picked. Both pickers fall back to a plain template field the moment you type {{ — useful when the value comes from an earlier node.
  • Resource fields (like a Fund picker on a finance node) work the same way — a searchable list of records with a template fallback.
  • Number and ID fields accept either a literal value or a template reference like {{event.task_id}}.
  • Enum fields with more than five options become a searchable list.
  • Date fields use a calendar picker.

Most node types ship their own structured form. Create, Update, Query, and Delete nodes for the records you work with — companies, employees, payrolls, advisers, initiatives, dashboards, documents, dynamic resources, attachments — open into purpose-built panels with the right input next to each field, rather than one long generic list. Create review task groups its inputs into Task content and Assignment cards. Query nodes show a Filters section with the right input per filter — a switch for booleans, a calendar for dates, a list for enums — so you can build a query without learning syntax.

Code fields. Some node inputs accept a small block of code (for example, the Code node's script body). These open in an inline editor with syntax highlighting and a Code assistant popover. Use the assistant to:

  • Format — clean up indentation and spacing without changing behaviour
  • Improve — rewrite for readability and fix obvious issues
  • Add comments — annotate what each part does
  • Simplify — shorten the code while keeping it equivalent

The assistant streams its rewrite back into the editor so you can accept or keep editing.

Issues. The Issues popover shows any problems Hub finds with your graph — dangling edges, missing required inputs, invalid expressions. A process can't be published until it has zero issues.

Draft vs. Publish. Hub auto-saves every change as a draft. Publish turns the process on so it starts running for real. A draft process won't fire its triggers. Click Unpublish to take a process offline without deleting it.

Export. From the toolbar menu, choose Export YAML to download the process definition. You can keep this as a backup, share it with another administrator, or import it back via the Import button on the Pipelines list.

Runs. Each time a process fires, it creates an execution you can review from the process's Runs panel.

Related