AI image generator
Small businesses need images constantly — social posts, product mockups, banners, thumbnails — and rarely have a designer available for each one. AI image generation covers a meaningful part of that gap.
It is included in the AI Workspace rather than being a separate subscription, with a monthly allowance on the Lite and Pro plans.
What it works well for
- Social media graphics — backgrounds, illustrations, seasonal posts.
- Concept mockups — showing a client an idea before committing to a shoot or a build.
- Blog and article images instead of generic stock photography.
- Backgrounds and textures for use behind text in banners and slides.
- Illustrations for explaining a process or service.
What it is still poor at
Being specific here saves wasted credits:
- Text inside images. Improving, but still unreliable. Add text afterwards in a design tool.
- Your actual products. It cannot generate a photo of a real item it has never seen. Photograph real products.
- Specific real people, including your team.
- Exact brand consistency across a series — colours and style drift between generations.
- Precise technical diagrams where accuracy matters more than appearance.
Writing prompts that produce usable images
Describe four things, in roughly this order:
- Subject — what is in the image.
- Style — photographic, flat illustration, 3D render, watercolour.
- Composition — close-up, wide shot, overhead, where the subject sits in frame.
- Lighting and mood — soft daylight, high contrast, warm, clean and minimal.
Then add what to exclude. “No text, no watermarks, no people” removes the three most common unwanted elements.
Specify aspect ratio for the destination — square for feed posts, wide for banners, tall for stories — rather than cropping a square afterwards and losing the composition.
Credits and cost
Image generation is more computationally expensive than text, which is why it draws on the premium credit allowance rather than being unlimited. Lite includes 30 images per month and Pro includes 60.
The practical way to make that go further: get the composition right in one or two low-effort attempts before generating final variations. Most wasted credits go on regenerating with vague prompts rather than on genuine iteration.
Rights and disclosure
Two things worth understanding before using generated images commercially.
First, the terms governing commercial use come from the underlying model provider and can differ between models. If an image is going on packaging or paid advertising, check the terms for the model you used.
Second, some platforms and advertising standards now expect AI-generated imagery to be disclosed in certain contexts. Where an image could be mistaken for a photograph of your actual premises, product or team, disclosure is the safer choice — and misrepresenting a generated image as a real one is a genuine consumer-protection risk.
Frequently asked questions
What is AI image generation good for in business?
Social graphics, concept mockups, article images, backgrounds and process illustrations. It is a strong substitute for generic stock photography.
Can it generate images of my actual products?
No. It cannot produce a photograph of a real item it has never seen. Photograph real products and use AI for concepts, backgrounds and illustrations.
Why does text inside AI images look wrong?
Text rendering is still unreliable even in improved models. Generate the image without text and add typography afterwards in a design tool.
How many images do I get?
Lite includes 30 per month and Pro includes 60, drawn from the premium credit allowance since image generation costs more to run than text.
Can I use AI-generated images commercially?
Commercial-use terms come from the underlying model provider and vary between models, so check them for anything going on packaging or paid advertising.
Do I need to disclose that an image is AI-generated?
Some platforms and advertising standards expect it in certain contexts. Where an image could be mistaken for a real photo of your premises, product or team, disclose it — presenting a generated image as real is a consumer-protection risk.