Here is what actually happens when a team adopts an AI image tool.
Eight concepts take 30 seconds to generate. Agreeing on one of them takes three days.
The outputs are scattered across everyone's downloads folder. Feedback moves through chat — "the second one from the left, in the third batch" — because someone has to describe coordinates out loud. And the prompt that produced the image exists nowhere. It's in one person's head, and when that person is out, nobody can reproduce even something close.
Generation got faster. The team didn't. The bottleneck didn't disappear; it moved from making to organizing and agreeing.
That is why XBRUSH ships team, folder, and history features alongside the agent rather than after it. They aren't two separate feature sets. They're two halves of one problem.
Half one — automation changes who can make things
The XBRUSH AI agent knows the generation and editing functions at the API level. Describe the result you want in plain language and the agent picks the tool, chooses the engine, writes the prompt the model needs, and runs it.
What that automation really changes isn't speed. It's who holds the ability to generate.
It used to be the designer, alone. Everyone else collected references, explained, and waited. Remove the cost of learning the tool and the strategist, the marketer, and the founder all draft their own first pass.
Which is exactly where the second problem starts. More people making things means more things made — and more things scattered. A tool that offers only automation stops right here.
Half two — infrastructure owns what happens next
Teams — output accumulates in the team, not in a person
The base unit in XBRUSH is a team, not a personal account. Each team carries its own members, billing, and content library.
Switch teams, create one, or open team settings from the top of the left control panel. Split teams by client and neither the outputs nor the credits get mixed up.
The important part is where things land. Generated results accumulate in the team's library, not in your downloads folder. When the person who made them leaves, the work stays.
Roles — let people in, and still keep control
Several people generating in one space does not mean anyone can do anything. There are four roles.
| Role | Manage members | Manage folders | Create content | View content | Billing | Delete team |
|---|---|---|---|---|---|---|
| Owner | ✅ | ✅ | ✅ | ✅ | ✅ | ✅ |
| Admin | ✅ | ✅ | ✅ | ✅ | ❌ | ❌ |
| Editor | ❌ | ❌ | ✅ | ✅ | ❌ | ❌ |
| Viewer | ❌ | ❌ | ❌ | ✅ | ❌ | ❌ |
There is one Owner per team, holding billing and deletion rights. Admins invite, remove, and change roles but cannot touch billing or delete the team. Editors create content and use team folders. Viewers are read-only.
Why four levels matters becomes obvious in practice.
- Viewer for clients and executives — they follow progress directly, without being able to delete anything or spend credits
- Editor for contract designers — they work in the team folders without seeing billing or restructuring the member list
- Admin for team leads — headcount management gets delegated; the payment method does not
Invitations go out by email address. Pick a role, send the invite, and the member shows as pending until they accept. Removing a member revokes access to the team and its content immediately.
Note: only the Owner can delete a team, and deletion permanently destroys all its content with no recovery.
Folders — it stays organized
Create folders under Home, then keep nesting subfolders as deep as you need. Drag an image from the results feed into a folder to file it. Select the team and you see everything; select a folder and you see only that folder.
Teams and folders can also be shared by email with other XBRUSH users, so you can open exactly one project and leave the rest closed.
Prompts and seeds — results become reproducible
This is the part teams feel most. Open any output and the right-hand panel carries:
- The full prompt — with a copy button
- Model, seed, and steps — every setting that produced the image
- Resolution and creation date
Fix the seed and you can reproduce the same result. Which means "one more in that tone from last week" becomes an actually executable request. The person who made it doesn't need to be at their desk — someone else copies the prompt and continues.
From that same detail view you go straight into Imitate, Upscale, Edit, Outpaint, Remove BG, or Animate. The place you review feedback is the place the next task starts.
Task comments — feedback lives next to the output
Design feedback has always accumulated somewhere other than the work: chat, email, meeting notes. So people describe coordinates out loud, and finding that conversation two months later is effectively impossible.
In XBRUSH, comments attach to the individual generation task. Open a task in the team results feed, click the comment icon, and every member of the team can read it and reply. Nobody leaves the platform.
When the output and the opinion sit in the same place, two things change. Nobody describes coordinates anymore, and the reason a version was chosen stays attached to the version.
Why neither half works alone
| What happens | |
|---|---|
| Agent only | You make things fast. But outputs scatter into personal folders, prompts vanish, and feedback piles up in another app. None of it becomes a team asset |
| Collaboration tools only | Filing and sharing work fine. But making things is still the job of the few people who know the tool. The bottleneck is untouched |
| Both, joined | More people make things, and what they make stays with the team. The increased volume becomes a library instead of chaos |
Automation widens participation; infrastructure absorbs it. With only one, the other becomes the bottleneck.
What it looks like as an actual workflow
Say a café is producing promotional images for a new menu item.
- The strategist (Editor) describes what they want and gets 8 concepts, then drags them into a "September Promo" folder
- The designer (Editor) opens the folder and comments on the one they like — "drop the background tone one step"
- Right there they open the detail view, copy the prompt, move into Edit, and produce the revision. The seed stays fixed
- The founder (Viewer) opens the folder to check progress. They cannot delete anything or spend credits by accident
- The approved cut goes straight from the detail view into Animate
Across those five steps, the number of trips to a messenger, a cloud drive, or a separate editor is zero. What got saved isn't generation time. It's the time spent moving between things.
Trying it before committing a team
If you're evaluating this for a team, start in Open Studio. You can watch other people's tasks without logging in, and once logged in you can generate using the shared credit pool — your own credits are never touched.
The generation panel is identical to the Workspace, so whatever you learn there transfers directly to your team's workspace.
Frequently Asked Questions
How do I invite teammates?
In team settings, click Add Member, enter the person's email address, and choose a role (Admin, Editor, or Viewer) before sending the invitation. They receive an invite email and appear as pending until they accept. Removing a member revokes their access to the team and its content immediately.
What are the team roles and how do they differ?
There are four: Owner, Admin, Editor, and Viewer. There is one Owner per team, holding every permission including billing and team deletion. Admins invite, remove, and change member roles but cannot manage billing or delete the team. Editors create content and use team folders. Viewers are read-only and can only view.
Can I see the prompt for an image later?
Yes. The output detail view shows the full prompt with a copy button, along with the model, seed, steps, resolution, and creation date. Because the seed is recorded, the same result can be reproduced — so a different teammate can pick up work they did not originally create.
Where does team feedback go?
Comments attach directly to individual generation tasks. Open a task in the team results feed and use the comment icon or the Add comment box; every team member can read the comment and reply to it. There is no need to move the discussion into a separate messenger.
Can I run more than one team?
Yes. Each team has its own member roles, billing, and content library. Click the team name at the top of the left control panel to open the team list, switch between teams, or create a new one. Splitting teams by client or project keeps outputs and credits from mixing.
In short
- Automation — the agent translates plain language into tool, engine, and prompt. Generating stops being the designer's job alone
- Teams — output accumulates in a team library, not a personal folder. Separate members, billing, and content per team
- Roles — Owner, Admin, Editor, Viewer: participation and control at the same time
- Folders — nested organization, with team- or folder-level sharing by email
- History — full prompt, model, and seed attached to every output, making reproduction and handover possible
- Comments — feedback stays next to the work, so no coordinates and no second app