> ## Documentation Index
> Fetch the complete documentation index at: https://docs.ideatura.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# AI Flows

> Distinguish AI-assisted execution from saved code.

This guide uses AI Flow to describe a workflow that calls AI during execution. The name does not imply a separate command or creation screen in your version of Studio. A saved code automation and a Skill serve different purposes: code performs an operation, while a Skill supplies instructions.

<h2 id="choose-the-approach">
  1\. Choose the approach
</h2>

Use a checked code automation for a repeatable operation with defined inputs. A flow may be appropriate when execution needs interpretation or decisions, but its outputs still need inspection. See [Context and instructions](/ai/context-window).

<h2 id="check-usage-before-running">
  2\. Check usage before running
</h2>

Do not infer that an AI Flow rerun is free because it was saved. AI work during execution can affect credits. Check the terms and usage shown for your account and compare authoring with execution separately in [Credits](/help/credits).

<h2 id="review-coverage">
  3\. Review coverage
</h2>

Begin with a small batch. Compare requested items with checked, skipped and failed items. If the flow stops or hits a limit, retain the exact error and identify what remains unknown. Breaking a task into smaller checks can make coverage easier to inspect; it does not guarantee lower total cost or faster execution.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.