Tutorials

Guided walkthroughs to build production workflows.

1. Trigger: Model Signals

Goal: Send a Welcome Email when a User model is created.

1. Define Trigger: json { "type": "signal", "config": { "model": "auth.User", "signal": "post_save", "condition": "created == true" // Only on create, not update } }
2. Add Action: Use the smtp.send or http.post connector (e.g., to SendGrid). json { "action": "http.post", "config": { "url": "https://api.sendgrid.com/v3/mail/send", "headers": { "Authorization": "Bearer {{$secret.SENDGRID_KEY}}" }, "json": { "personalizations": [{"to": [{"email": "{{event.instance.email}}"}]}], "subject": "Welcome!" } } }

2. Trigger: Inbound Webhook (Stripe)

Goal: Provision access when a checkout.session.completed webhook arrives.

1. Configure Ingestion: Set up a generic webhook endpoint /api/automate/webhooks/stripe/. Configure signature verification middleware in settings.py.
2. Define Rule: json { "and": [ { "==": [{ "var": "type" }, "checkout.session.completed"] }, { "==": [{ "var": "data.object.payment_status" }, "paid"] } ] }
3. Add Action: Call your internal API to provision access.

3. Human-in-the-loop Approval

Goal: Require admin approval for refunds > $500.

1. Step 1: Check value. json { "if": [ { ">": [{ "var": "amount" }, 500] }, "needs_review", "auto_approve" ] }
2. Step 2 (Branch: needs_review): Send Slack notification with "Approve" button.
3. Step 3: core.wait_for_signal (Pause execution). The workflow suspends until a specific callback signal is received from the Slack interaction.
4. Step 4: Process Refund.

4. LLM Chaining

Goal: Summarize support ticket and suggest reply.

1. Step 1: llm.chat.
   • Prompt: "Summarize this ticket: {{event.payload.description}}"
   • Model: gpt-4o
2. Step 2: llm.chat.
   • Prompt: "Draft a polite reply based on this summary: {{step.1.output.text}}"
   • JSON Schema: Enforce output structure {"subject": "...", "body": "..."}.
3. Step 3: Save draft to Zendesk/Freshdesk via HTTP.