> For the complete documentation index, see [llms.txt](https://help.smartico.ai/welcome/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://help.smartico.ai/welcome/products/vibe-studio-beta/working-with-the-ai-assistant.md).
# Working with the AI assistant
The assistant is the heart of Vibe Studio. You describe what you want in plain English; it reads the whole project, makes the changes across all the files, and the preview refreshes. Your job is to direct it, not to code.
### How it works
Type a request in the chat and press **Enter** (Shift+Enter for a new line). The assistant explores the project, edits the files, and checks the result; you watch its steps stream in. Bigger requests take longer.
### If you're new to AI-assisted building
If this is your first time building with an AI assistant, a few habits will make the whole experience much smoother. The assistant is fast and capable, but think of it as a talented junior teammate: excellent at first drafts, and at its best with clear direction and a careful review.
* **Always test - don't take "done" on trust.** The assistant will happily say something is finished and working. Check it yourself: preview as a real (test) player, on both desktop and mobile, and click through the actual flow. Run `/verify` (and `/widget-check` for widgets) before you publish.
* **Test the tricky cases, not just the happy path.** Try a player with no data, a very long name, an empty state, and a different language. That's where rough edges hide.
* **Easy to start, harder to finish.** Your first result can look impressive within minutes - but polishing it into a production-ready version usually takes several rounds. That's completely normal; plan for iteration rather than expecting it to be perfect in one shot.
* **Go one step at a time.** Small, specific requests are easier to achieve and to check than one giant brief. Change one thing, look at it, then move to the next.
* **Be concrete.** "A countdown to Friday 8pm in red" is better than "add a timer". Say what content, what style, what behavior.
* **Save versions as you go.** Before a bold change, save a **Version**. If a few attempts break too much, roll back to the last good one instead of trying to unwind them.
* **If the assistant gets stuck**, it often helps to `/clear` , and describe the goal fresh, `/undo` back to a working state, or rephrase with an example or a screenshot. A clean context frequently unblocks it.
* **Publishing isn't the finish line.** After you embed the widget or share the link, check it live on a real device too.
* **Stuck for real? Ask us.** If you can't get past something, Smartico support is here to help - reach out to your Success Manager.
### Show it what you mean
* **Attach an image** with the **+** button, paste one directly, or drag and drop files into the chat. Use a screenshot or a mockup and say "reproduce this", "use these colors", or "match this layout".
* **Built from Figma.** Attach a Figma frame URL and the assistant builds from your design.
* **Generate images.** Ask it to create an icon, an illustration, or a background, and it produces and places them for you.
* **Point with Inspect.** Turn on **Annotate**, click an element in the preview to pin a comment on it ("make this bigger", "wrong color here"), then send all the pinned comments at once. Pointing beats describing positions in words.
### It remembers the project
The assistant keeps **project memory** - a summary of what each file is for, and the decisions it made - and re-reads it every turn, so it stays consistent across a long session. You can see it in the **Memory** tab, and refresh it with `/memory-review`.
### Keep usage efficient
Each label has a monthly AI allowance. The context bar (top-right of the tab bar) shows how full the conversation is. When you switch to a new task, run **`/clear`** - it starts a fresh conversation while keeping your project memory, so the assistant runs faster and uses fewer tokens. See Slash commands.
### FAQ
**It changed the wrong thing.** Run `/undo` to revert the last change, then re-ask more specifically - or use Inspect to point at the exact element. You can use `/redo` to return the last change back
**It's taking a long time.** Complex requests do a lot under the hood. Watch the Trace tab; if you want it to stop, press the stop button in the composer.
**The assistant says it's done, but something's still wrong.** Don't take "done" on trust. Preview as a test player on desktop and mobile, click through the real flow, and run `/verify` and `/widget-check` (for widgets). If something's off, point it out - use **Annotate** to click the exact element - and ask for the fix.
**Can I just edit the code myself?** Yes - the code editor is there - but most changes are faster to ask for.
**I'm stuck - it keeps missing what I want.** Try `/clear`, and describe the goal from scratch, `/undo` back to a working state, or add a screenshot or example to guide it. If you're still stuck, Smartico support can help - reach out to your Success Manager.
---
# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.
## 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, and the optional `goal` query parameter:
```
GET https://help.smartico.ai/welcome/products/vibe-studio-beta/working-with-the-ai-assistant.md?ask=&goal=
```
`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal.
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.