You generate an image in an AI tool. The text is wrong. You re-prompt. The composition shifts. You re-prompt again. The colours change. You re-prompt a third time. The text is still wrong and now the background is different too.
That loop, the one every creative has been stuck in since AI image generation started, is what Google Pics is trying to kill.
Google just launched Pics at I/O 2026, an AI image generator and editor built directly inside Google Workspace. It is powered by Nano Banana 2, Google’s latest image model, and it does something no mainstream tool does well: let you click on any part of a generated image and change just that part without touching the rest.
What Pics Actually Does
Pics generates images from a text prompt, same as ChatGPT or Midjourney. The difference starts after generation. Every element in the image is treated as a separate object. Click on the cat. Type “change this to a dog.” Pics changes the cat. The background, the text, the layout, the colours, all unchanged. Just the cat becomes a dog.
Three capabilities define the tool.
Object-Level Editing
Click anything in the image. Move it. Resize it. Remove it. Swap it. Transform it. The surrounding area reconstructs itself to fill the gap or adjust to the new position. No re-prompting. No starting over. The image stays intact while you change individual pieces of it.
Text Editing Inside Images
Click on any text in a generated image and modify it directly. Change the words, fix a typo, translate it to another language. The font, size, and style stay the same. For anyone making flyers, invites, posters, or social cards inside Google Slides, this alone eliminates the “re-roll and hope” problem.
Import Your Own Photos
Pics is not limited to generated content. Upload your own images and use the same editing tools on them. Click on an object in your photo. Remove it. Move it. Resize it. The model fills in the background behind where it was. No Photoshop subscription required.
How It Works Inside Workspace
Pics launches as its own web app first, available to Google’s trusted testers starting today. It rolls out more broadly this summer for Workspace Business Standard and higher, plus Google AI Pro and Ultra subscribers.
The first Workspace integration point is Google Slides. Generate and edit images without leaving your presentation. Everything saves straight to Drive. The plan is for Pics to embed into more Workspace apps over time, though Google has not confirmed Docs or Gmail yet.
Pricing is not separate. Pics will be included in Workspace Business Standard, Business Plus, Enterprise, and Google AI Pro and Ultra plans. No standalone option announced.
The Workflow That Changes Everything
Here is the before and after that matters.
Before Pics: Generate an image in ChatGPT. Notice the text is wrong. Re-prompt. The composition shifts. Re-prompt again. The colours change. Re-prompt a third time. The text is still wrong and the background is different now too. Give up and take it to Canva for manual editing. Export. Import into Slides.
After Pics: Generate inside Slides. Click the text. Fix it. Click the object you want to swap. Describe the replacement. Drag elements to new positions. Done. No app switching. No re-rolling. No starting over.
The editing loop collapses from “generate, review, regenerate, review, regenerate, give up” to “generate, point, describe, done.”
What to Look Out For
The demo Google showed had a noticeable quirk. When the presenter asked Pics to change a cat to a dog, the resulting dog still looked quite cat-like. Object replacement fidelity will improve, but expect some weird hybrids in the early days.
SynthID watermarking is enabled by default on all Pics output, along with C2PA Content Credentials. Every image Pics generates is traceable back to the model. Good for provenance. Worth knowing if you are generating images for client work that needs to look non-AI.
The biggest open question is how well object editing holds up on complex images with lots of overlapping elements. Clean single-object swaps on simple backgrounds look convincing in demos. Cluttered compositions with intertwined elements are where these tools tend to struggle. Watch for that as real-world tests come in.
Also worth noting: Pics enters a crowded field. Canva has been building AI editing into its design suite for over a year. ChatGPT’s image generation is fast and improving. Adobe Firefly has generative fill. Pics differentiates on the object-level precision and the Workspace integration, but the feature gap will narrow fast.
Why This Matters
The creative tool landscape has been converging on the same problem from different angles. Canva gives you manual control over elements. ChatGPT gives you fast generation. Midjourney gives you quality. Adobe gives you professional editing. Pics is the first tool that combines generation with object-level editing in the same surface, and it is doing it inside the productivity suite where most business design work already happens.
For small business owners creating social posts, flyers, and presentations, the practical impact is simple: fewer apps, fewer exports, fewer rounds of regeneration. You generate once and edit in place. The image becomes a living document, not a static artifact you keep re-rolling.
That is the shift worth watching. Not the model quality, which will keep improving everywhere. Not the pricing, which will converge. The workflow. Pics makes the image editable in a way that matches how people actually think about design: “I like this, but change that one thing.” That is how design conversations work in real life. Now the tool works the same way.
Want more like this? Join MOKU CLUB for free. Weekly resources, early access to new guides, and occasional templates you can actually use. Join below.
