For a few years now, making an image with a large language model (LLM) has followed a familiar rhythm.
Users need to describe a scene they want to be generated, wait for the AI to work, and fingers crossed only to get something close. And when something does come pretty close, one detail is wrong: a cup sits too far left, a hand is misshapen, a headline uses the wrong city name.
The usual fix is to rewrite the whole request and generate again, which often changes the parts that were already fine.
People who do this for work have started treating that loop as a tax on the job rather than a quirk of a new tool.
That habit is what Google is naming on the product page for 'Pics.'
Available as a Workspace image app that began a wider rollout on September 1, 2026 after a first look at I/O in May, the company frames the old method as trial and error prompting.
Pics is meant to sit after the first generation, so a person can point at one object or one line of type and ask for a local change instead of rolling the entire frame.
Pics is both a web app and an overlay inside Workspace.
In Docs and Slides, users can select an existing picture and open the same editor without leaving the file. Drive support for opening almost any saved image in Pics is scheduled in the coming weeks.
Files can also come from Google Photos or a local upload. Creations live in Drive, can be shared by link, and can be edited with more than one person on the same canvas, which is the part that most clearly belongs to Workspace rather than to a chat window that happens to draw pictures.
The model underneath is Nano Banana, used together with Gemini.
Official feature lists are consistent across the Workspace blog, the consumer blog, and the admin rollout note.
Users generate from a prompt and usually get several variations. They can isolate an object and move, resize, or replace it, or leave a short comment on that region the way they might comment in a Doc.
Text inside the image can be rewritten, restyled, or translated while the surrounding layout is supposed to stay put.
There are crops for web, social, print, and digital formats, upscaling to 2K or 4K, batching of several edits at once, and a version history so a bad change can be undone.
Official videos walk through social posts, in-Slides mockups, Drive drag-and-drop brand work, shared canvases, and the generate-then-refine loop.
Access is not universal.
Pics is listed for Workspace Business Standard and Plus, Enterprise Standard and Plus, Google AI Pro and Ultra, Google AI Pro for Education, and AI Expanded Access.
Rapid Release domains started seeing it on September 1, with visibility taking up to 15 days.
Scheduled Release domains start September 15 on the same gradual clock. Generative use is capped.
Google says limits are higher at least through February 28, 2027, after which access may tighten with notice. Edits started from Slides still count against existing image-generation limits on those surfaces.
The product sits in a crowded Google catalog, which showed up immediately in replies to the Workspace announcement.
People already have Gemini image tools, Imagen, Nano Banana as a model name, image features in Docs and Slides, and Google Photos edits.
Pics is the Workspace-native generator-plus-editor with object selection and in-image type, not a replacement for Photos and not the same thing as a one-shot image in a Gemini chat.
Outside Google, the comparison that keeps coming up is Canva or Adobe Express.
Those products still let you build a layout from templates, stock, and manual tools, with AI as an option. Pics starts from a prompt or an uploaded file and then offers local edits.
Tech coverage has noted the missing marketplace of creator templates and the fact that you are still asking a model to draw, not placing type on a blank board yourself.
Early reporting from the I/O demos already hinted at the gap between the slogan and the output.
In one walkthrough, changing a cat on a party invite into a dog left an animal that still read as mostly cat. That is the practical version of the problem Pics is trying to shrink: local edits that do not quietly rewrite lighting, style, or nearby objects.
Business teams who need a flyer in two languages, a product dropped onto a new background, or a slide figure that is almost right will be the first to find out how often the rest of the picture stays still.
The tool is in the same apps those teams already open. The remaining question is whether pointing at a detail is enough, or whether the old regenerate-and-hope cycle just moves one step later in the same workflow.
What Pics really wants to change, is the cost of being almost right. A slide or a one-page handout that is 90% usable no longer has to be thrown out because of a single object or a single line of type.
The file stays in Drive, the history stays attached, and a colleague can open the same canvas instead of inheriting a flattened export from another app. That is a workflow argument more than a quality argument, and it is the part of the launch that does not depend on the slogan.
The naming still gets in the way of that simpler story. Nano Banana is the model. Pics is the app. Gemini still makes images. Photos still edits photos. Older product names have not left the conversation.
For an audience that was promised fewer steps, an extra product identity is another thing to look up before anyone clicks an object and types a note. Replies to the Workspace post treated that overlap as the first reaction, which is a useful signal.
People noticed the catalog before they noticed the editor.
