SupportGet started

Agent

Describe what you want, and the agent writes the prompts, picks the model and makes it.

Last verified

Overview

The agent is a chat that makes images for you. Reach for it when you don't know what you want yet, or when you're tired of writing prompts by hand. Tell it what the work is for, and it writes the prompts, picks the model and checks the results.

In a picture book project, "Give me 6 different directions for the cover of this book" got six ideas to tap; the agent knew the book from the project's Instructions. After two more taps, four covers appeared in the chat and in the feed.

Try asking

  • "Make six matching app icons for a weather app: sun, rain, snow, wind, fog and storm." One prompt per icon, run at once, in one style.
  • "Find the four Fern covers I made today, remove their backgrounds, then make them stickers." It searches your library, then runs both edits in order.
  • "Which of these four is strongest for the cover? Upscale that one." It compares them, says why, and upscales the winner.

What it can do

  • Make images, up to 16 prompts in one request.
  • Offer directions to choose from, or ask one question when a wrong guess would waste work.
  • Edit one image, "the second one" or a whole stack: remove backgrounds, upscale, Ideogram Edit, restyle, recolor, stickers and slices.
  • Make a video, starting from one of your images.
  • Find earlier work in your library by what you asked for, the model, the kind of job or the day, and save images to a collection.
  • Favorite or delete images, create projects, and set up a Batch table for you to run. Deleting always asks first.

Start a chat

Select Agent at the bottom right of the page. On a phone, it's the Agent tab. An empty chat opens on "What are we making?" with four cards to start from.

To bring in an image from the feed or the lightbox, choose Ask agent.

Show it images

Select + or type @ to upload, pick from your library or add a collection, or drop files on the chat. A message carries up to 8 images.

Say what each image is for and how they combine. Given the Halvard bar and the lynx sticker with one sentence about the sheet, Nano Banana 2 made one sticker sheet with the lynx holding a square of chocolate.

A collection stays on for every image in the chat until you remove it, up to 5 per chat.

While the agent works, keep typing. Your message waits above the box until you press Steer. Stop or Esc ends the run.

Pick the model

The model chip starts on Auto, and the agent picks a model for each request. For the book covers it chose Nano Banana Pro; for the flat icons, Nano Banana 2.

Pin a model from the chip instead. If it's a poor fit, say for lettering, the agent suggests a better one under the result, with a button to try it.

Projects and teammates

In a project, the agent reads its Instructions and collections before every reply. It calls the Instructions "the brief", and when you settle on a direction it can update them, with an Undo.

Chats belong to the workspace. Anyone in it can read and continue any chat.

Skills

A skill is a set of instructions for one kind of work, such as print, social posts or sprite sheets. Type / or choose Browse skills under +. Ask the agent to save how it did something, and it becomes a skill of your own.

What it costs

Talking to the agent is free. What it makes costs the same as on the Create page: the sticker chat above paid 2 credits per cutout, and the stickers were free. A big request waits for you to press Generate.

Tips

Fill in the project's Instructions first. The agent reads them every time, so "this book" is enough.