Single-step “email me when X” flows are fine, but Activepieces earns its keep on chains: trigger, transform, branch, loop, notify. These ten multi-step workflows are the ones that pay rent on a business server, and each one starts as a short spec you dictate to the vibe-code agent in the chat. Describe the outcome; it builds the flow, tests it, and leaves you a draft to review and publish.

1. Order-to-fulfillment chain

The moment a paid order lands, three things must happen: record it, start fulfillment, tell a human. The chain: WooCommerce trigger → add a row to the orders Google Sheet → branch on shipping method → email the packing list to the warehouse → post to your ops Slack channel. Type: “When an order is paid in WooCommerce, log it to our ‘Orders’ sheet, email a packing list to warehouse@company.com, and post the order number and items to #fulfillment on Slack.” The agent wires the trigger, maps the line items into both messages, and tests with a real or sample order before you publish.

2. Lead capture with dedupe

Website form → CRM → sales alert, with a dedupe step so the same person doesn’t become four leads. Webhook trigger → Code piece checks the email against the leads sheet → if new, add the row, create the HubSpot contact, notify #sales; if existing, update the row and stop. Type: “When our contact form submits, check if the email is already in the ‘Leads’ sheet; if not, add the lead, create a HubSpot contact, and ping #sales. If it exists, just note the new inquiry on the existing row.” Good dedupe is the difference between a pipeline and a junk drawer — and it only exists because the flow checks before it writes.

3. The daily digest, not fifty pings

Per-event notifications train people to mute channels. Collect instead: Schedule trigger at 8:50am → read the last 24 hours of form entries, sign-ups, or orders → Code piece formats a summary → one email or Slack message at 9am. Type: “Every weekday at 9am, send me one Slack message listing yesterday’s new leads, orders, and failed logins, with counts and links to the sheets.” You read one message with your coffee; the raw events stay in the sheets for anyone who wants detail. If the volume ever justifies per-event pings again, the same pieces rearrange in minutes.

4. Polite invoice chasing, with an escape hatch

Cash-flow automation that isn’t robotic: Schedule trigger reads overdue invoices from a sheet → branch on days overdue → send a tiered reminder sequence → after three emails with no payment, hand the list to a human instead of sending a fourth bot email. Type: “Every morning, check the ‘Invoices’ sheet for unpaid invoices, send a reminder based on how many days overdue each one is, and after three reminders send me a Slack message with the list so I can call them.” Tone is a business decision — the agent drafts the reminder emails, you approve the wording once, and the flow sends them forever.

5. Triage the support inbox

Support email arrives → classify → route → acknowledge. Gmail trigger on the support inbox → Code or AI piece classifies the message (billing, bug, other) → Router piece sends each type to its own Slack channel → auto-reply to the customer with expectations. Type: “When a support email arrives, classify it as billing, bug, or other; post billing to #finance-support, bugs to #dev, and reply to the customer with our support hours and a ticket reference.” Set expectations honestly in the auto-reply — a fast “we got it, reply within 4 business hours” beats silence, and beats an overpromised one-hour SLA even more.

6. Airtable ↔ Google Sheets sync

The team lives in Sheets, the ops person lives in Airtable, and someone has been copying rows by hand. Two flows fix it: on Airtable record update → upsert the Sheets row; on Sheets row change → update the Airtable record. The Store piece keeps last-write state so the two flows don’t loop forever fighting each other. Type: “Keep the ‘Campaigns’ Airtable base and the ‘Campaigns’ Google Sheet in sync both ways — never create duplicates, and log any conflicts for me to resolve.” Two-way sync is genuinely fiddly; let the agent build the loop guard and explain how it breaks ties, so you can reason about it later.

7. Approvals for anything risky

Some actions shouldn’t fire on their own: refunds over $200, discount codes, mass emails. Add the Approval piece as a human-in-the-loop step — the flow pauses, sends you an approve/reject link, and only continues on approval, with the decision recorded in the run. Type: “When a refund request comes in over $200, pause the flow and send me an approve/reject link in Slack; only process it if I approve, and log my decision in the sheet.” This is automation with a brake pedal: you keep the speed, lose the cliff, and every override leaves an audit trail.

8. Onboarding checklist for new customers

New signup → create their world: form trigger → add a row to your Customers table → create their folder or account in your other apps via the HTTP piece or the Google Workspace pieces → send the welcome email with next steps → notify the account manager in Slack. Type: “When someone completes our onboarding form, add them to the ‘Customers’ sheet, create a folder for them in our shared Drive, send the welcome email from our template, and tell their account manager in #accounts.” The checklist lives in the flow now, not in someone’s head — and it runs the same way on a Sunday signup as a Tuesday one.

9. Content pipeline with a human gate

Publish on schedule without publishing by accident: Schedule trigger → pull the latest draft from your RSS feed or content sheet → Code piece formats the post → Approval step sends you a preview → on approve, post via the LinkedIn or Twitter piece. Type: “Every Monday at 8am, draft a LinkedIn post from our latest blog article, send me a preview in Slack to approve or reject, and only post it if I approve.” One bad auto-post costs more trust than a hundred scheduled ones earn — the gate is the entire point of this flow.

10. A watchdog for your flows themselves

Flows fail silently: an API key expires, a sheet gets renamed, and nobody notices for two weeks. The watchdog closes that loop: Schedule trigger → query the Activepieces API for recent failed runs → Code piece formats the failures → alert you in Slack and by email. Type: “Every morning, check for flows that failed in the last 24 hours and send me one Slack message listing each flow, the step that failed, and the error text.” The platform that automates everything else gets to automate being watched — failures become a morning note instead of a customer complaint.

Pick the one that costs you hours

Choose the workflow that eats the most time this month and dictate its spec — the agent builds it as a draft, tests it with sample data, and you publish when it looks right. If you’d rather have the whole setup done for you, this is the shortcut: Activepieces on OpenSysLab.