# Tips & Best Practices

Small habits make the AI assistant dramatically faster and more accurate. These are the practices that consistently produce better results.

## Write Better Prompts

**Be specific.** Replace *"make it look better"* with *"change the server nodes to blue and arrange them in a row."* The AI isn't guessing your intent — it's using the words you give it.

**Include counts and names.** *"Add three API gateway nodes"* beats *"add some API gateways."*

**State what to do, not what not to do.** *"Use dashed lines for internal connections"* works better than *"don't use solid lines for internal connections."*

## Be Precise About What to Change

**Use @mentions** to pinpoint exactly which elements you want to modify, especially in complex diagrams. See [@Mentions & Context](/models/work-in-the-app/ai/mentions.md).

**Select first, then ask.** Select the elements on the canvas and use `@Selected` in your message — often faster than typing each name.

## Always Review Before Applying

The preview is there for a reason. Check it carefully before committing, especially for:

* **Large changes** (10+ elements) — use the expanded full-screen preview.
* **Renames or rewires** of existing elements — it's easy to miss a mis-wired link in a thumbnail.
* **Anything on a shared board** — others see what you commit.

## Reposition Before You Commit

After you **Merge** or **Add**, the staged changes show on the canvas with a dashed green boundary — drag the whole boundary around to place the new content exactly where you want it before committing.

## Use the Right Tool for the Job

| Task                                      | Use                                                         |
| ----------------------------------------- | ----------------------------------------------------------- |
| Fix typos or translate labels             | [AI Text Adjust](/models/work-in-the-app/ai/text-adjust.md) |
| Group related nodes                       | [Cluster by Keyword](/models/work-in-the-app/ai/cluster.md) |
| Create or modify shapes and connections   | [AI Chat](/models/work-in-the-app/ai/chat.md)               |
| Ask a question about your ServiceNow data | [AI Chat](/models/work-in-the-app/ai/chat.md)               |
| Recreate a hand-drawn diagram             | [Attaching Images](/models/work-in-the-app/ai/images.md)    |

## Iterate, Don't Restart

You don't need to get the perfect result on the first try. Reply with corrections:

* *"Move the database node a bit to the right."*
* *"Change the link color to green."*
* *"Rename the top-left shape to 'API Gateway'."*

The AI keeps the context of the current session, so short follow-ups work well.

## Start from Suggestions When Stuck

If you're not sure what to ask, open a new chat and click one of the quick-start suggestion cards. They're good examples of the kinds of prompts the AI handles well.

## Keep Sessions Focused

Use separate sessions for separate lines of work. Unrelated context can confuse the AI and slow down responses. See [Managing Chat Sessions](/models/work-in-the-app/ai/sessions.md).

## Respect the Preview Flow

Don't skip the preview, even if it feels faster. The green staged area is also your undo — once you commit, undoing a large AI change manually is painful.

## Related

* [AI Chat](/models/work-in-the-app/ai/chat.md)
* [@Mentions & Context](/models/work-in-the-app/ai/mentions.md)
* [Reviewing & Staging Changes](/models/work-in-the-app/ai/staging.md)


---

# Agent Instructions: Querying This Documentation

If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://docs.ins-pi.com/models/work-in-the-app/ai/tips.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.