> ## Documentation Index
> Fetch the complete documentation index at: https://docs.ideaboxai.com/llms.txt
> Use this file to discover all available pages before exploring further.
# Working with Automations
> A step-by-step walkthrough of creating, editing, and monitoring an automation.
This page walks through creating, editing, and monitoring an automation end to end. For a conceptual introduction, see Automations Overview.
## Getting started
Go to **Automations** in the left navigation and click **Create**.
Describe the workflow you want, or pick an example prompt to start from.
Check the trigger, the steps, and where the output goes.
Turn it on, or keep refining with Persona AI until it's ready.
Keep an eye on the **Automation Overview** dashboard to track runs and catch anything that needs attention.
## Creating an automation
Automations live in their own section in the left-hand navigation. To start one, click **Create** and you'll land on a blank canvas with a single prompt: *"What would you like to automate?"*
You can describe a workflow in your own words, or start from one of the suggested examples, such as:
* *"When a new application is submitted, evaluate eligibility using credit score, LTV, and DTI."*
* *"Every morning, surface all pending applications with aging and risk status for review."*
* *"When an application ages beyond 30 days, alert the underwriter for immediate attention."*
## From prompt to running workflow
Once you describe what you want, Persona AI breaks the request into steps, picks the right connectors and knowledge bases, and builds the automation for you, including a plain-language explanation of what it created and why.
In the example above, the following prompt produced a two-step workflow:
*"Every weekday at 1:45 PM, pull today's approved and pending applications from the Underwriting Pipeline dashboard, summarize them, and email the summary to the loan committee"*
1. **Query today's applications** — pulls approved and pending mortgage applications from a connected knowledge base, including status, risk tier, credit score, LTV, and DTI.
2. **Create summary and email results** — generates a breakdown by priority tier and SLA status, then sends it to the loan committee over Outlook.
New automations are saved as a **draft** first. Persona AI will ask whether to activate it right away or make changes before it goes live. Nothing runs on a schedule until you confirm.
## Editing and managing steps
Once an automation is active, every part of it stays editable. You can adjust the trigger, add new steps, or fine-tune an individual step without rebuilding the workflow from scratch.
From the workflow detail page you can:
| Action | Description |
| --------------------------------------- | ------------------------------------------------------------------------------------------------ |
| **Reconfigure the trigger** | Change the schedule (e.g., from weekly to monthly), or switch to a manual-only trigger. |
| **Add a step** | Just ask Persona AI in plain language, and it inserts the new step in place. |
| **Inspect any step** | See exactly what it did, what data it used, and what it produced. |
| **Re-run a single step** | Or the whole automation on demand with **Run now**. |
| **Change connections** | Swap out or add tools like Outlook, Jira, or a connected knowledge base. |
| **Switch between List and Canvas view** | List gives you a step-by-step breakdown; Canvas gives a visual map of the trigger and each step. |
Because everything is conversational, corrections are just as easy as the original request; asking Persona AI to change the email updates that field directly, no form-hunting required.
## Monitoring your automations
The **Automation Overview** dashboard is your control tower for everything you've built. It shows totals at a glance, along with recent outputs and anything that needs a second look.
From here you can see:
* **Total automations**: how many are active, paused, or still in draft
* **Needs attention**: a live count of failed runs or steps waiting on human input, so problems don't get buried
* **Recent outputs**: the latest results across all your automations, with a direct link to view the full output
* **All automations**: a searchable, filterable list (by status, trigger type, or date created), switchable between List and Card layouts
When everything is healthy, the Needs Attention panel simply confirms it: *"All clear — nothing needs attention."*
## Run history and outputs
Every run, scheduled or manual, is logged with its status, execution time per step, and the final output. If a run succeeds, you can open the output directly from the workflow or from the Recent Outputs list on the dashboard. If something fails or needs a human decision partway through, it surfaces in Needs Attention rather than failing silently.
Automations is under active development, and a few rough edges are still being worked out. None of these affect the core flow of describing, running, and monitoring a workflow; but if something looks off, it's worth checking whether it's a known Beta issue before assuming it's a configuration mistake.