Ask the AI Assistant in plain language.
Write what to create or change in the AI dock below the canvas, and the assistant builds and edits objects, scenes, motion, and interactions for you. One undo takes back a whole request.
Ask the AI
The AI Assistant is a beta feature. You need to be signed in, and each request uses your account’s AI credits. If the toolbar has no AI button, see When you can’t use AI first.
-
Open the AI dock
Click the sparkle-shaped AI button at the right end of the toolbar at the top center, past the divider. Or press ⌘K. The AI dock opens at the bottom center of the canvas. The cursor goes in the input field that reads “Describe what to create or change.” The row below it has the Text and Image tabs and the up-arrow Run button.
Clicking the AI button at the end of the toolbar opens the AI dock at the bottom center of the canvas. Video: Clicking the sparkle-shaped AI button at the right end of the toolbar opens the AI dock at the bottom center of the canvas. A soft five-color light circles the input field, with three examples above it. -
Select what to change
To change objects that already exist, select them on the canvas first. The bottom row of the AI dock shows how many you selected, like “Objects: 2 Selected.” Now type “Make this blue,” and the selection changes. To make something new, you don’t need to select anything.
Selecting a circle and a rectangle shows “Objects: 2 Selected” in the bottom row of the AI dock. Video: With the AI dock open, clicking a circle and then Shift-clicking a rectangle shows Objects: 2 Selected in the dock’s bottom row, and the examples change to fit the selection. -
Type a request and run it
Type what to create or change in the input field, then press ↩. To start a new line, press ⇧↩. The first time you use AI in a document, you can also click an example above the input field to fill it in. The first time you send a request to AI, a “Before you use AI” note appears in the chat before the request goes out. Read it, then click Agree and continue to run the request. While the assistant works, the elapsed time shows, like “Working… 3s.”
Clicking the example “Add a title and a subtitle” and pressing ↩ adds a title and a subtitle a moment later. Video: Clicking Add a title and a subtitle above the input field fills it in. Pressing Enter shows a Working line counting seconds, then a title and subtitle appear in the scene and glow pink. -
Watch a large request arrive in parts
A large request that needs more than about 25 changes is handled by the AI in parts, one after another. That happens when you make several scenes or many cards at once. As each part finishes, its result lands on the canvas first. Meanwhile the chat shows which part it’s on, like “Working on part 2… 12s.” Once the last part is done, the assistant replies once for the whole request. One request runs to at most six parts. To stop partway, click the × Cancel button where the run button was. The parts that already landed stay.
The title and the first row of cards land first and glow, then “Working on part 2” shows and the second row of cards lands. Video: A request for intro cards for eight team members is sent with Enter. The title and the first row of four cards arrive first and glow, then “Working on part 2…” shows and the second row of four arrives. -
Check the result
When it’s done, the assistant replies with one line about what it did, and a change receipt such as “2 changes” appears below. New or changed objects glow briefly. If something new is off screen, the view moves to it. Hover over the change receipt to make them glow again, and click it to select those objects.
Hovering over “2 changes” makes the title and subtitle glow, and clicking it selects both. Video: Hovering over 2 changes below the reply makes the title and subtitle glow pink again, and clicking it selects both. -
Undo, or keep refining
If you don’t like it, click Undo next to the change receipt. Everything that request changed goes back at once. For small fixes, keep typing, like “Make the title bigger.” The assistant also sees the last 3 text requests in this conversation that ran to completion, along with its replies.
Clicking Undo on the change receipt removes the title and subtitle together. Video: Clicking Undo next to the change receipt removes the new title and subtitle together, and that Undo button dims.
Undo on a change receipt works only when the request is your last action
If you edit something yourself after a request, that button dims. That’s because you’d have to undo the edits after it first, in order. In that case, press ⌘Z to undo one step at a time. For a request handled in parts, each part is one step.
What you can ask for
| What | Examples and what it does |
|---|---|
| Create | “Add a title and a subtitle” · “Lay out three photo frames in a row.” Creates text, shapes, frames, and tables in the current scene. |
| Edit | “Make this blue” · “Round the corners.” Changes color, size, position, font, and effects. Select what to change first. |
| Scenes | “Add a thank-you scene at the end.” Adds, duplicates, reorders, and resizes scenes. In other scenes it sees only object names and types, so open the scene you want changed before you ask. |
| Motion | “Make everything come in one by one” · “Transition smoothly to the next scene.” Adds transitions between scenes and timeline keyframes. If auto-keyframe is on in Animation mode (it’s on at first), position, size, and opacity changes become keyframes at the playhead. |
| Interactions | “A button that goes to scene 3 when clicked” · “A quiz that keeps score.” Writes rules that respond to clicks or key presses. For requests that fetch values from the web, build them yourself. Requests and variables shows how. |
| Variables | “Make the price a variable and use it in three places.” Creates variables and links them to text. |
| Questions | “What’s in this scene?” · “What did you just change?” It only answers, and changes nothing. |
| Images | Make new pictures in the Image tab. The Text tab only uses images already in the project. See Make an image below. |
Make an image
Depending on your account, image generation may not be available. Then clicking Image shows a note in the row above the input field that this AI feature isn’t available, and the run button turns off.
-
Choose the Image tab
Click Image in the bottom row of the AI dock. A row for choosing the size, aspect ratio, and format appears below the input field.
Clicking Image brings up the size, aspect, and format row below the input field. Video: Clicking Image in the bottom row of the AI dock brings up the size, aspect, and format row and Use selected image below the input field, and the examples change to picture descriptions. -
Choose a size and aspect ratio
Choose a size and aspect ratio in that row. Size · 2K is about twice as wide and tall as Size · 1K, and uses more credits. Aspect ratios are values like Aspect · Auto or Aspect · 16:9. The format starts at Format · WebP quality. To keep the image exactly as received, choose Format · Original. Your choices stay the same next time.
-
Choose a base image
To build on an existing image, select a single image on the canvas and turn on Use selected image. Only PNG, JPEG, and WebP images work. To draw something new, skip this step.
Selecting a photo makes Use selected image available, and clicking it turns it on. Video: In the Image tab, the dimmed Use selected image becomes available after clicking the photo on the canvas, and clicking it turns on the check box. -
Describe it and run
Type the picture you want, or what to change, and press ↩. The new picture lands in the center of the current scene, sized to fit inside it. It’s named “AI Image,” and the base image you used stays as it is. One undo takes this back too.
Typing “A soft spring background” and pressing ↩ places a picture in the center of the scene a moment later. Video: Typing A soft spring background in the Image tab and pressing Enter places a picture in the center of the scene a moment later. It glows pink, and the reply Added the image appears.
Credits and usage
AI requests use your account’s AI credits. Credits have a monthly limit and a daily limit. If your account has limits, the bottom row of the AI dock shows how much you’ve used, like “12 / 300 · resets Nov 1,” and the ring around the run button fills as you use them.
The meter shows whichever of the two limits has less left. For the daily limit, a time shows instead of a date. At 80%, the ring turns yellow. If you don’t see this meter, your account has no limit.
Every request the assistant answers uses credits. A request handled in parts uses credits for each part, and one request runs to at most six parts. Questions that only got an answer, and requests that ended with “Nothing to apply” or “That change was too big to finish in one go,” use credits too. Images use far more credits than text requests, and Size · 2K and Use selected image use more than Size · 1K. If a request needs more than your remaining credits, no credits are used, and you only get a “You’ve used…” reply.
Create with AI uses the same credits, for each scene the AI designs, and that use counts on the same meter.
When you can’t use AI
When AI can’t be used, the reason shows in the row above the input field, and the run button turns off. When a single request isn’t accepted, the reason stays in the conversation as a reply. Either way, the request you sent comes back to the input field, so you can edit it and send it again.
| What you see | What it means and what to do |
|---|---|
| No AI button on the toolbar | Your account can’t use AI, or AI is paused for now. In that case, pressing ⌘K does nothing either. If AI is only paused, the button comes back on its own when AI is turned on again. |
| “Sign in to use AI” | You aren’t signed in. Click Sign in next to it. In the desktop and tablet apps, a browser opens. Once you sign in and come back to the editor, the AI dock checks again. |
| A note that this AI feature isn’t available | Your account can’t use this feature. If it shows only in the Image tab, only image generation is unavailable to your account. |
| “Verify your email to use AI.” | Depending on your account, you need a verified email to use AI. Once you verify your email, you can use it right away. |
| “You’ve used today’s AI credits.” · “You’ve used all your AI credits for this month.” | You’ve hit the limit, or this request needs more credits than you have left. Requests that use lots of credits, like images, can get this reply even when the usage meter shows credits left. Switch to a text request or a Size · 1K image, or send it after your credits refill at the time or date in the reply. |
| “Too many requests” | Apart from credits, the number of requests has a per-minute limit and a daily limit. Send it again after the number of seconds in the reply. If you hit the daily limit, you may need to wait a few hours. |
| “Nothing to apply” | The assistant didn’t change anything. The request comes back to the input field. Describe in more detail what to change and how, then send it again. |
| “That change was too big to finish in one go.” | The AI’s answer ran too long and was cut off. Ask in smaller pieces, one scene or a few objects at a time. |
| “The request was too big to finish here…” | The request went as far as one request can go (up to six parts) and stopped. What it changed so far stays. Ask it to go on. |
| “Stopped at part 3: …” · “Cancelled at part 3.” | The run stopped and kept what the earlier parts changed. The reason follows the colon, and “Cancelled” means Cancel was clicked. Ask again for what is left, or click Undo on the change receipt to take back the whole request. |
| “You’re offline” | Your internet connection dropped. Reconnect, then click Retry next to it. |
| “AI is temporarily unavailable.” | Something went wrong on the AI side for a moment. Send it again in a few minutes. If there’s a Retry next to it, click it. |
| “AI features are paused right now.” | An operator paused AI for now. A note from the operator may show instead of this message. If it shows in only one of the Text and Image tabs, only that feature is paused. While the AI dock is open, it checks again every minute. When AI is back on, the run button turns on by itself. |
| “AI features are turned off for this account. …” | AI is no longer available for this account. If you think this is a mistake, contact support. |
| “This document is too large for the assistant …” | There’s too much document information to send. Despite what the message says, it has nothing to do with selecting scenes. Usually the current scene has too many objects. Open a scene with fewer objects before you ask, or split the big scene in two. |
| “Editing waits while AI makes this project.” | Create with AI is making scenes in this window. Until it’s done, the input field and the run button are off, and requests waiting their turn are cleared. Wait until it’s done, or click Stop creating on the progress line in the AI dock. |
What’s sent and what’s kept
When you send a text request, what you wrote, the last few turns of the current conversation, the document’s structure, and your current selection go through SSHOW’s servers to the AI provider. For the current scene, that includes object names, positions, text, and styles. For other scenes, only the names and types of the scenes and their objects are sent. An image request sends only what you wrote, plus a scaled-down copy of the base image if you use one.
The conversation on screen is kept only in the browser or app you’re using, up to the last 30 turns per document. The record belongs to the account that made it, so if another account opens the document in the same browser or app, the conversation starts empty. When that account sends a request, the earlier account’s record is erased. SSHOW’s servers keep what you wrote and usage records (count, time, and cost) for up to 400 days. These records don’t include document content, images, or what the AI made. The AI provider may keep records for a set period (for example, 30 days) for safety checks. Your content isn’t used to train AI models.
The “Before you use AI” note that appears in the chat with your first AI request briefly says what is sent and where. Click Agree and continue and your account keeps the record, so you aren’t asked again until the AI Usage Notice changes. Click Cancel and the request isn’t sent; it stays in the input field. Your account agrees once, so other AI features such as Create with AI don’t ask again.
For details, see the i-shaped What is sent? button in the bottom row of the AI dock, or the AI Usage Notice.
Working with the AI dock
You can send the next request while you wait. While the assistant works, type your next request and press ↩. It waits its turn. When the input field is empty, the run button becomes an × Cancel button that stops the current work.
Start the conversation over. Click New chat in the row above the input field. The conversation so far stays, dimmed, and the assistant no longer refers to it. To erase it completely, click Clear in the row below the dimmed conversation.
You can move the AI dock. Drag an empty spot around the input field to float the dock anywhere over the canvas. Drop it near the bottom, and it docks again. Double-click an empty spot to send it back to the bottom center.
Shortcuts on this page
| Open or close the AI dock | ⌘K |
|---|---|
| Run the request | ↩ |
| Start a new line | ⇧↩ |
| Bring back a request you sent before | ↑or↓ |
| Undo, when the input field is empty | ⌘Z |
| Close the AI dock from the input field | esc |
Updated 10/14/2026 · The videos on this page were recorded in the current editor.