Convert to Workflow

After an AI agent completes a task, you can convert the successful run into a reusable workflow. The workflow replays the same steps without needing the AI model - making it faster and cheaper.

How It Works

The AI agent records every action it takes during a session
When the session completes, you can convert it to a workflow
The conversion creates workflow nodes for each navigation, click, type, wait, and extraction step
Self-healing: if the agent retried an extraction, the final successful version is used
The workflow can then be scheduled and run repeatedly

Via the Dashboard

Run an AI agent session and wait for completion
Click Convert to Workflow on the results page
Review and optionally edit the generated workflow
Save and schedule if needed

Via the API

POST https://api.scrayle.com/api/v1/ai/agent/{jobId}/convert

Headers:
  x-api-key: YOUR_API_KEY

Response:
{
  "workflowId": "wf_abc123",
  "name": "Converted from AI Agent - job_xyz",
  "nodeCount": 8
}

What Gets Converted

The conversion translates AI agent actions into workflow nodes:

Agent Action	Workflow Node
Navigate to URL	`navigate_to_url`
Click element	`click_element`
Type text	`type_text`
Wait for element	`wait_for_element`
Extract data	`extract_data`
Execute code	`execute_code`
Scroll page	`scroll_page`

💡Self-Healing

If the AI agent retried an extraction (e.g., because a selector failed on the first attempt), the conversion uses the final successful extraction, not the initial broken one.

Benefits

Faster execution - No AI model inference needed, runs in milliseconds per step
Lower cost - Workflow runs cost 5 credits (vs. potentially hundreds for AI agent)
Schedulable - Run the workflow on a cron schedule
Editable - Modify individual steps in the visual workflow builder
Reliable - Deterministic execution of the same steps every time