Automate your first workflow
Turn a recurring question into a scheduled workflow that runs on its own. You can build the automation conversationally inside Claude, or directly in the Rig app, whichever fits how you work.
In Claude
Describe the workflow to Claude with Rig connected. Claude proposes a Rig automation, the data sources, transformations, and outputs, and you confirm or refine before it ships.
In Rig
🧱Whether you build an automation on your own or use Claude to prompt one, Rig automations are a visual canvas: chain nodes together so data flows step-by-step.
Create a new automation
- Automations > New Automation.
- Choose a template or Blank Canvas.
- Pick a trigger:
- Manual
- Schedule (hourly, daily, weekly, biweekly, monthly, timezone-aware)
- Query Check (only runs when query returns rows)
- Webhook (HTTP POST)
Node types (cheat sheet)
| Node | What it's for |
|---|---|
| Deterministic Query | Explicit, reproducible data pulls (auditable numbers). |
| Iterator | Run logic once per row (e.g. account-by-account summaries). |
| Agentic Researcher | Exploration and synthesis when you're not sure which tables to use. |
| Reporter | Turns gathered data into a readable narrative with a prompt file. |
| Integrator | Push to Slack, email, HubSpot, Notion or Chargebee. |
| Skill | Run a playbook defined in Claude or Cursor (MCP client) as part of the workflow. |
| Internal Query | Query Rig's own context layer and metadata graph. |
Send results to Slack
A scheduled workflow can post its result to a Slack channel. It posts through the same Slack app as @rig, so a workspace owner has to install that first (see Ask Rig in Slack), and the app has to be a member of the channel. Invite it before the first run. Rig checks membership before it posts, and otherwise the step fails with Slack post failed: The Rig bot isn't a member of Slack channel '…' — invite it to the channel, then retry. For a private channel the bot was never invited to, the message says the channel isn't in this Slack workspace or the bot hasn't been invited to it yet.
A Slack post that fails does not stop the run. The rest of the workflow carries on, and the run finishes with an amber completed with errors badge whose tooltip shows the Slack error, so look for that badge after the first scheduled run as well as for a red failed one. Only two set-up problems stop the run: a Channel ID that refers to an earlier step returning more than one row, and a step Rig cannot run at all, such as one of a type it no longer supports. Every other Slack failure, including the bot not being in the channel, ends as completed with errors.
- Give the workflow a Schedule trigger.
- Add an Integrator node after the step whose output you want to send, and set Tool to Slack - Send message. Slack Action is then Send Message.
- In Channel ID, paste the channel's ID, which starts with C (some older private channels start with G). To find it in Slack, click the channel name at the top of the channel and scroll to the bottom of the details window.
- Leave Workspace ID blank. Blank uses your workspace's own Slack install, which is the one set up for @rig.
- Write the Message. Insert a field from an earlier step as
@node_1.field, as the hint above the box shows, and it is posted in full. A bare@node_1is shortened: a step that returns data posts only its row count, long structured output becomes [structured output not shown], and other text is cut at 3,000 characters. Whenever something is held back, Rig usually adds a Full report: link to the end of the message. - Tick Only send on scheduled runs so that manual and test runs skip the post and nobody gets a message while you are still building.
A channel shared with another organisation through Slack Connect is refused until an owner opts it in, in the same way as for @rig.