In CloudRadial AutomationAI, AI Generate can start from a workflow you already have and build an updated version from a plain-language description of the changes. This article covers picking the workflow, describing what should change, reviewing the expanded prompt, and putting the updated workflow into service. It is for the Admin and Owner roles.
Important -> AI Generate saves a new copy and never changes the original workflow. The original keeps running until you switch its triggers over to the new copy in Step 4.
We'll follow four main steps:
- Opening AI Generate From an Existing Workflow
- Describing What Should Change
- Reviewing the Expanded Prompt
- Testing and Switching Over
Step 1: Opening AI Generate From an Existing Workflow
Why are we doing this? Starting from the existing workflow gives AI Generate its current steps, so you only describe what should change.
Open Workflows and select the row actions menu next to the workflow you want to update. Select AI Generate from this…. The AI Generate Workflow panel opens with that workflow already selected under Start from.
You can also select AI Generate at the top of the Workflows page and pick the workflow under Start from.
Step 2: Describing What Should Change
Why are we doing this? AI Generate builds from your description, so the more specific it is, the closer the draft will be to what you want.
Complete the fields in the panel:
- Name: the name of the new workflow. It defaults to the original name followed by "(copy)"
- Folder: where the new workflow is filed. Select New folder to create one
- Steps: how the steps are built: Agent steps, PowerShell scripts, Python scripts or C# scripts
- Start from: the workflow you are updating
- What should change: the change in plain language. Name the systems involved, the order of actions and where results should go
For example: "After adding the CC, add an internal note to the ticket listing the email address that was added and who requested it."
Select Expand prompt.
Step 3: Reviewing the Expanded Prompt
Why are we doing this? AutomationAI checks your description against your installed extensions before it builds anything, so you can correct it first.
The panel switches to an editable Expanded prompt box. It starts with your description, then lists the steps resolved against your installed extensions, including the extension, the tool and the parameters each step will use. Below the box, the panel lists the extensions it plans to use and confirms that the original workflow is not changed.
Correct anything that is wrong, such as a parameter marked "not specified" that you can fill in. Select Back to change your description, or select AI Generate to build the workflow. AI Generate builds the workflow from exactly the text in the box.
The button reads Generating… while AutomationAI drafts and validates the workflow. This can take a minute or more, and generation counts toward your workspace's Usage Credits.
When generation finishes, the new workflow opens in the Designer and the AI Generation panel shows a summary of what it built. Check the steps on the canvas and open individual nodes to confirm their configuration.
Step 4: Testing and Switching Over
Why are we doing this? The new copy is a draft, and the original keeps running until you replace it.
In the Designer, select Test draft to run the new workflow. When it works as expected, publish a version and deploy it to the runner that holds the secrets it needs.
Then move the original's triggers to the new workflow:
- Webhook: the new workflow has its own webhook URL and secret. Open Properties, select Enable webhook and copy the URL and secret. The change saves to the draft and takes effect when you publish, so publish first, then update every form or system that calls the original webhook with the new values
- Schedules: open Properties on the new workflow and add the same schedules under Routines that the original uses. Active schedules run the deployed version
Once the new workflow is running in production, retire the original's deployment. See Editing and retiring deployments.
Making Small Changes Without a Copy
If the change is limited to the script in one node, you can edit it in place in the Designer with the inline AI assistant instead of generating a new workflow. See Writing node scripts with the inline AI assistant.
Next: With the updated workflow deployed and its triggers moved, review its first runs in History.
If you are still having trouble, we're here to help! Submit a ticket here and check our status page for any service interruptions.
Comments
0 comments
Article is closed for comments.