Skip to main content
Workflows let you chain a trigger (something happens) to one or more actions (things Hiveku does in response). Use them to send Slack notifications when a form is submitted, email a daily report every morning, welcome new signups, or sync data between systems.

​
Create Your First Workflow

1

Open Workflows

Click Workflows in the top navigation. Workflows are account-scoped, so they’re shared across every project.
2

New Workflow

Click New Workflow, give it a descriptive name (e.g., Slack on form submit), and open the visual editor.
3

Add a trigger

Every workflow starts with one trigger node. The most common:
  • Webhook — fires when an HTTP POST hits a unique URL
  • Schedule — cron-style timing (e.g., every day at 9 AM)
  • Database change — row inserted, updated, or deleted
  • New visitor — someone lands on a page
  • Form submission — any form on your sites is filled out (spam and duplicates are filtered out before the trigger fires)
  • SMS received — an inbound text arrives, optionally scoped to one number or keyword
There are also event triggers for CRM, bookings, helpdesk tickets, billing (invoices, payments, subscriptions), and surveys — see the Workflow Node Reference.
4

Add action nodes

Click Add Node (or drag from the palette) and connect nodes with edges. There are 350+ nodes across categories like CRM, email, SMS, social, CMS, AI image and video generation, web research, voice, accounting, and deploys — browse them in the Workflow Node Reference. The staples:
  • Send email — transactional or custom HTML
  • HTTP request — POST/GET to any URL
  • Database query — read or write any table
  • AI Agent — route work to a department agent with your brand context, or a bare model for quick classification
  • Delay — wait N seconds/minutes/hours/days
  • Notification — Slack, Discord, in-app
5

Reference earlier data with merge variables

In any text field, type {{ to autocomplete a variable like {{trigger.email}} or {{ai.response}} — inserted references render as pills that show whether they resolve. If a variable can’t be resolved when the workflow runs, the value is dropped rather than sent as literal {{...}} text, and the miss is flagged on that step in the run inspector.
6

Add conditions (optional)

Drop in if/else nodes to branch based on data. For example: if lead score > 50, notify sales; else, add to nurture list.
7

Save and enable

Click Save, then flip the Enabled toggle. Hiveku validates the workflow first — structural problems block enabling with a message naming the broken node, while missing optional configuration just asks you to confirm. Once enabled, the workflow is live.

​
Example Recipes

Form submission to Slack

Form submission trigger -> Send notification to Slack channel -> Add to CRM contacts

Daily report

Schedule (9 AM daily) -> Database query -> AI generation (summary) -> Send email

Welcome sequence

Database change (users table insert) -> Delay 5 min -> Send email -> Delay 2 days -> Send email

Order confirmation

Database change (orders table) -> Send email -> HTTP request to shipping API

​
Run History

Click any workflow to see its run history:
  • Completed — finished without errors
  • Failed — one of the steps errored
  • Running — currently in flight (useful for long delays)
  • Pending — queued but not yet started
Click any run to drill in: input data, step-by-step state at each node, and the exact error message if it failed.

​
Dead Letter Queue

When a run fails, it lands in the Dead Letter Queue for that workflow. From there you can:
  • Replay — re-run with the original input after fixing the root cause
  • Dismiss — mark as handled without re-running
Don’t replay blindly. Hiveku skips emails, texts, and posts the original run already completed, but if the failed run reached other side effects — charging a customer, writing to an external API — confirm the failure point before replaying.

​
Automatic Safeguards

Enabled workflows protect themselves:
  • Circuit breaker — 5 consecutive failed triggered runs pause the workflow automatically. Runs you start by hand (Manual Runs, Test Runs, replays) never count, so editing a half-configured node can’t pause a workflow whose real traffic is fine.
  • You’re told, not just paused — account owners and admins get an email naming the paused workflow and why, with a resume link. The workflow list also shows the paused state with the number of trigger events blocked while paused.
  • Resume catches up — events that arrived during the pause are recorded and replayed automatically when you resume, oldest first. Fix the failing step, click Resume, and the backlog drains itself.
  • Daily AI budget — each workflow can execute at most 200 AI-agent node runs per day (and 25 within one run). Crossing the ceiling pauses the workflow and emails the owners, so a mis-wired loop can’t burn AI credit overnight.

​
Versions

Every time you save a workflow, Hiveku snapshots a version. Open Versions in the top right of the editor to see the history, diff two versions, or roll back instantly. Great for recovering from a bad edit.

​
Credentials

Shared credentials (Slack webhook URLs, API tokens, database passwords) live in Workflows > Credentials. They’re stored encrypted and referenced by name — so you update a key in one place and every workflow using it picks up the new value.

​
Verifying It Works

In the editor, click Test Run and provide sample input. Watch each node light up as it executes and inspect the output. Once Test Run passes, enable the workflow and trigger it for real.

​
Troubleshooting

  • Webhook — confirm you’re POSTing to the exact URL shown (regenerate it if unsure), and that the workflow is enabled
  • Schedule — confirm the cron expression is valid and that the timezone matches what you expect
  • Database change — confirm the trigger is on the right table and operation (insert vs update)
Most action failures are credential issues: expired tokens, revoked access. Open the run, click the failed step, and read the error. Update the credential in Workflows > Credentials and replay.
Fix the underlying cause first (credential, bad data, downstream API outage) and then replay. Replaying without a fix just produces the same failure.
Check the trigger — a Database change trigger on a table you update inside the same workflow can loop. Add a condition that skips updates originating from the workflow itself, or use a last_updated_by column.
Check your email — auto-pause always notifies account owners and admins with the reason (repeated failures, a detected loop, or the daily AI budget). Open the newest failed run to find the failing step, fix it, then click Resume. Events that arrived while paused replay automatically.
An unresolved merge variable is dropped rather than sent as literal {{...}} text. Open the run and look for the “merge variables resolved to nothing” notice on the step — it lists the exact paths that missed. Usually the fix is a typo in the variable path; use the {{ autocomplete to insert the correct one.

​
What’s Next?

Workflows Reference

Trigger types, merge variables, safeguards, and advanced patterns

Workflow Node Reference

All 350+ nodes by category — AI, CRM, CMS, creative, voice, and more

Send Emails

Pair workflows with transactional email for automated outreach