AI Workflow Generation: Build Automations from Plain English
AI workflow generation is a Krom Automation Pro feature that turns a plain-English description of an automation into a complete, ready-to-review workflow. Instead of manually adding a trigger, chaining actions, and configuring each field, you describe what you want and Krom Automation builds it for you.
It’s the fastest way to get a new workflow running, especially for users who are still learning how triggers and actions connect.
How it works #
You type a description of the automation you want. Krom Automation’s AI analyses that description, maps it to the available triggers and actions from both the free and Pro plugins, generates a complete workflow configuration, and saves it as a new workflow on your site.
The generated workflow starts in Paused status. You review it on the canvas before activating, so you can adjust any field, add conditions, or change delays before anything runs on real data.
The AI has full awareness of every trigger and action Krom Automation supports, including all Pro integrations. It can generate workflows that combine multiple integrations in a single automation.
How to generate a workflow #
- Go to Krom Automation > Workflows in your WordPress admin.
- Click the Generate with AI button (separate from the standard Add New Workflow button).
- A prompt field opens. Type a description of the automation you want to build.
- Click Generate.
- Krom Automation sends your description to the AI, processes the response, and creates the workflow.
- The new workflow opens on the canvas in Paused status, ready for you to review.
Screenshot: Workflows list page with the Generate with AI button highlighted.
Screenshot: AI prompt input field with an example description typed in.
Screenshot: Generated workflow open on the canvas in Paused status, showing the trigger and action nodes the AI created.
Writing a good prompt #
The quality of the generated workflow depends on how clearly you describe what you want. A few things that consistently produce better results:
Name the trigger event explicitly. Don’t say “when someone buys something” — say “when a WooCommerce order is completed.” The more specific the trigger, the more accurately Krom Automation maps it.
Name the integrations you want to use. If you want the workflow to add the customer to Mailchimp, say Mailchimp specifically. Krom Automation knows all 24 Pro integrations by name.
Describe the outcome, not the steps. “Send a welcome email to the customer and subscribe them to my Mailchimp audience” is better than “add a Send Email action and then add a Mailchimp Subscribe action.” Let the AI handle the step selection.
Include timing if it matters. If you want a delayed response, mention it: “Send a review request email 7 days after order completion.” Krom Automation will add the appropriate delay to the trigger.
Examples of effective prompts:
- “When a WooCommerce order is completed, send a thank-you email to the customer and subscribe them to my Mailchimp post-purchase audience.”
- “When a LearnDash course is completed, send a congratulations email, tag the student in ConvertKit, and post an alert in my Slack #courses channel.”
- “When a new user registers, wait 10 minutes then send a personalised welcome email using AI to generate the message body.”
- “When a MemberPress payment fails, send a payment failure email to the customer and notify the admin via Slack.”
Reviewing and adjusting the generated workflow #
Always review a generated workflow before activating it. The AI produces a solid starting structure, but there are things only you can fill in:
API-specific IDs. The AI can create a Mailchimp Subscribe action, but it can’t know your Mailchimp Audience ID. Open any integration action nodes and confirm the IDs, list names, or channel names are set correctly for your accounts.
Email content. The AI populates subject lines and message bodies with placeholder text. Replace them with your actual copy, or use the AI Generate Text action inside the workflow to produce dynamic content.
Merge tags. Check that the merge tags the AI used in action fields match what the trigger actually exposes. Run a simulation after reviewing to confirm they all resolve correctly.
Conditions. If the workflow needs conditional logic (for example, only send the Slack alert for orders over $200), add condition nodes manually after the AI generates the base structure.
Run control. Check the Run Control setting. The AI sets a sensible default, but confirm whether it should be Run once per entity or Run every time for your specific use case.
Running a simulation after generation #
Before activating any AI-generated workflow, run the simulator.
Click Simulate in the toolbar. The simulator runs the workflow against real data from your site and shows you exactly what each step would do, what merge tags would resolve to, and which steps would fail. This catches any configuration gaps the AI left that you need to fill in.
See Testing Workflows with the Simulator for a full walkthrough of reading simulation results.
What AI workflow generation does not replace #
AI workflow generation is a starting point, not a finished product. It builds the skeleton of a workflow based on your description. You still need to:
- Enter your own API credentials and integration-specific IDs
- Write or approve the email content
- Test the workflow with the simulator before activating
- Monitor the first few real executions in the Logs to confirm everything works as expected
Think of it the same way you’d think of installing a template from the library. The template installs the structure; you configure the details. AI workflow generation does the same thing, but from a description you write rather than a pre-built template you pick.
Frequently asked questions #
Which AI provider does workflow generation use?
It uses whichever AI provider and API key you’ve configured in Krom Automation > Settings > AI Settings. The same provider that powers the AI Generate Text, Content Moderation, and Auto-Tag actions.
Can I generate a workflow that uses Pro integrations I haven’t configured yet?
Yes. Krom Automation generates the workflow based on the triggers and actions available, regardless of whether you’ve connected the integration. You’ll need to complete the integration setup before the workflow can actually execute those steps.
Is there a limit on how many workflows I can generate?
There’s no Krom Automation-imposed limit. Each generation uses your AI provider’s API, so it counts against your provider’s usage quota the same as any other AI action.
The generated workflow doesn’t match what I described. What should I do?
Rephrase your description and generate again. More specific language about the trigger event and integration names consistently produces more accurate results. You can also generate a workflow as a starting point and then edit the canvas manually to get exactly what you need.