PHOTOCRAFT AI
Generate an image.
Then keep editing it.
Two ways to work. The image tools turn a prompt or reference into a picture. The Agent works on the document you have open — renaming layers, changing opacity, editing text, adding brightness or black & white adjustments, and inserting generated images. Everything lands in the same editor you already know.
Agent and image tools
IMAGE TOOLS
Prompt or reference in, image out
Choose a model, aspect ratio and resolution. You see the exact credit quote before you confirm, and generation failures release their reservation. Reference-image editing is supported on models that offer it.
AGENT
Ask your editor for the change
Supported actions include creating and renaming layers, changing opacity, editing text layers, adding brightness or black & white adjustment layers, and inserting generated images. The Agent operates your PhotoCraft document — not Photoshop — and every proposed command is applied by your browser with an undo available.
How a task runs
- 01
Send the request.
The task is quoted, then accepted with a spending cap for Agent work or a fixed quote for images.
- 02
Watch plain-language progress.
The panel shows what is happening — preparing, processing, applying changes, saving — plus how long you have waited. No jargon states, no fake progress bars.
- 03
Review, undo, export.
Applied edits can be undone from the conversation. Generated images can be inserted into the document or downloaded. Export stays in your hands.
If a task cannot finish, unused reserved credits are released. Confirmed model usage stays charged; the billing rules are in the Terms of Service.
What it does — and does not — do today
- Supported: the document actions listed above, image generation and reference-guided editing on the enabled model catalog, one credit balance across both tools.
- Not supported: every possible editing request — the Agent operates a fixed set of document actions and will say when a change is out of scope.
- Privacy boundary: manual editing runs on your device. Using AI sends your prompt and the context or reference images you explicitly submit for that request; scope, providers and retention are in the Privacy Policy.
- Models: availability is shown in the editor before you confirm. The model name and settings are always displayed; no silent substitution.
Before your first AI task
Can I keep using the editor for free?
Yes. Manual editing is free. Optional AI tasks spend credits and show a quote before confirmation.
Does AI upload my whole document?
AI sends the prompt and the context or reference images you explicitly submit. Review the Privacy Policy before sharing sensitive work.
What if generation fails or I undo an edit?
Failed image generation releases its reservation. Confirmed Agent usage can remain charged, including when you undo an applied edit. Save your document before leaving.
Are example prompts guaranteed to work?
No. An example request is a starting point, not a promised output. Model availability, task limits and supported document actions are shown in the editor.
Learn it step by step
Start with the walkthrough, then go deeper in the guides: