Images for LinkedIn-style work-feed posts with AI: calm, clear, one idea each
Make calm images for work-feed posts with AI: one idea per picture, carousel covers with room for your headline, no handshakes, no logos, prompts to paste.
A work feed is a quiet room. People scroll it between meetings, with the sound off, and they are not looking for a spectacle. They stop for a post that looks like it was made by someone who knows what they are talking about. So the picture has one job: look calm, look specific, and get out of the way of the sentence.
That is a different brief from a photo grid or a Story, and it is where AI images most often go wrong: a glowing brain, a robot shaking hands, a ladder to a sunrise. This post is about the opposite approach, a small set of rules for pictures that suit a post about a lesson, a result or a how-to, written for the platform as a plain place where you post, with nothing here tied to any one company.
Rule one: a picture says one thing
Before you write a prompt, finish this sentence: "This post is about ___." If the blank needs an "and", you have two posts. A post about what you learned from a slow launch is about patience; it does not need a clock, a chart, a mountain and a plant. Pick one object or one place that carries the feeling and render only that.
- A lesson gets a quiet metaphor: one path, one door, one lit window.
- A result gets a calm scene of arrival, not a trophy: a finished table, a tidy desk, a bridge crossed.
- A how-to gets a clean, flat setup of the thing itself: a notebook, a tool, a tidy stack of cards.
- A question to your network gets an open space: a wide horizon, an empty chair at a long table.
Aerial view of a single narrow wooden jetty reaching into a still lake at dawn, soft mist, pale pink light, very large area of empty water and sky, calm and minimal, no people, no textRule two: leave the space on purpose, and typeset the words yourself
Image models are not reliable with text. A headline rendered inside the picture will come out with a swapped letter often enough that you cannot trust it, and a typo on a professional post costs more than a plain picture ever saves. So do not ask for words. Ask for room, then add the headline in the post editor or any layout tool you like, where every letter stays exact and you can change a word in seconds.
Say where the room goes. "Large empty area in the upper left" works better than "space for text", because the model has a place to put it.
A tidy wooden desk seen from above, a closed notebook and a single pen on the right, soft morning light from the left, large empty smooth area on the left half, muted warm neutrals, no text, no screensRule three: cover slides for document-style posts
Many work-feed posts are a swipeable document: a cover, a handful of slides, a closing line. The cover is the only slide most people will see before they decide, so it is the one to spend time on. Make it plain and consistent with the rest: one calm picture, large empty area, your headline set on top.
- Render the cover first, with the empty area in the same spot you will use on every slide.
- Write the style words once (surface, light, palette) and paste them into every other slide's prompt. For the full method of holding a set together, see how to make a carousel that looks like one set.
- For the inner slides, use fewer pictures than you think. A text slide with a flat colour taken from the cover often reads better than a new image.
- Make the last slide quiet. A calm picture and a single line is a good place to end.
Rule four: the shape of the frame
Work feeds tend to reward a picture that is easy to read on a phone without cropping surprises, and a portrait or square frame usually does that. Which ratios each platform shows, and what each is good for, is in the formats cheat sheet, so check there rather than here, and check your first post on a real phone before you batch more. The ratios a model can make are listed on its form in the studio; a quick, low-priced model such as P-Image is a good place to try a composition before you commit.
What to avoid, and why
- The handshake. AI hands are the first thing to break, and a handshake is two of them. It is also a stock-photo cliche. Show what the deal made possible instead.
- The glowing brain and the chrome robot. They say "AI" and nothing about your post.
- The staged meeting room. Smiling strangers around a laptop were never your team, and readers can tell.
- Real people. Do not generate someone who looks like a named person, and do not use a colleague's likeness without asking. A scene with no people avoids the whole question.
- Fake client logos or screenshots. An invented logo wall or dashboard implies customers or numbers you do not have. Leave out anything that claims a result the picture cannot prove.
One lit window in a quiet brick building at blue hour, the rest of the facade in soft shadow, calm and still, empty sky above, photorealistic, no people, no text, no signsA small routine that fits a working week
You do not need a batch for this. Write the post first, decide the single idea, render two or three options, keep one. If the first picture is wrong, change one word (the light, the place) rather than rewriting the whole prompt. A first top-up of €3.00 is the smallest you can make, and failed renders go back to the balance on their own, so trying three options for one post is not a risky habit.
If a picture is almost right but has one distracting detail, an edit model such as P-Image Edit can remove it without starting over. For planning several posts at once, the monthly content calendar shows how to batch by theme, and the quote card post covers one-line text backgrounds.
What the studio will not do for you
It makes the picture and stops there. It does not write your post, set your headline, check a claim or publish anything. It cannot show a real product, a real office or a real person accurately; a render is an illustration, and it should be read as one. If a post is about a real result, put the real evidence in the text, and let the picture set the mood.
Questions
What kind of AI image suits a professional post?
A calm picture with one idea: a single object or place that carries the feeling of the post, with an empty area for your headline. Skip glowing brains, robots and staged handshakes.
Can the studio put my headline in the image?
It is better not to. Image models often misspell text, so ask for an empty area and set the headline yourself in the post editor or a layout tool, where it stays exact.
Is it safe to use AI images of people in a work post?
Safer to avoid them. Do not generate a likeness of a real or named person, and do not invent clients, logos or screenshots. A scene or an object tells the story without those risks.
Which image shape should I use for a document-style post?
A portrait or square cover is a common choice. The formats cheat sheet lists the shapes and what each suits, and it is worth checking your first post on a phone.