Free AI social media post generator
Turn one idea into three publishable posts, written for the platform you pick. Enter the idea, choose Instagram, TikTok, LinkedIn, Facebook or X, and each draft comes back with its angle named, its caption shaped for that surface, and the hashtags that suit it. It takes about five seconds, and nothing here sits behind an account.
Free forever No login No credit card
This is as far as we can get without your material.
Paste any one of these and regenerate:
Want the ad that goes with it?
Copy any of the three, they are yours. Selzee reads your brand first and then makes the creative to go with copy like this, so the next batch comes from your own product rather than from a box you filled in.
Make the creative, freeThat is the free batch for this page. Reload for another set, or Start free and let Selzee post from your own brand.
An AI social media post generator turns one idea into a caption you can publish. This one is free, runs without a login, and writes for the platform you choose rather than producing one caption and relabelling it. Type the idea, pick where it is going and what it is for, and three drafts come back: each with its angle named, a caption shaped for that surface, and hashtags that suit it.
It is built for the people who actually post: founders, small marketing teams, agencies and anyone who has decided to start posting and keeps stalling at the caption. Rhythm is what the platforms reward, and the median post earns very little engagement on any of them, as Socialinsider's benchmarks across millions of posts show, so the goal is not one perfect caption. It is enough good posts, often enough, to find out what works for you.
Type the thing you want to post about: a change you made, a question customers keep asking, a product going live, a lesson from last week. One sentence is enough. This is the only field the generator cannot do without, because everything else it can infer and the idea it cannot.
The platform decides the shape of the caption, not just its length. A first line that earns a read on LinkedIn is a line Instagram hides behind a more link, and a thought that fits Facebook comfortably will not fit a single post on X. The goal matters just as much: a post written to start comments reads nothing like a post written to move someone to a link, and asking for both in one caption is how you get a caption that does neither.
Brand, audience, tone and any real product facts are optional, and the last one decides how much you get back: with real detail you get three posts built on three different pieces of it, with none you get one and a list of what would change that. Paste the actual objection a customer raised, the real feature name, the phrase your buyers use. The generator will not invent a statistic, a price or a review quote you did not give it, so the specifics in the output are the specifics you supplied.
Three drafts come back, each with its angle named, its caption written for the platform you picked, and its hashtags. Copy any of them with one button. Nothing sits behind a signup: the first batch runs the moment you click, with no account, no email and no credit card.
Same idea, different artefact. Every one runs without a login and nothing you make in them costs anything.
The hub for the seven. One idea into posts, captions, carousels, scripts or a week of ideas.
Try it free 📸Describe the photo and get five captions at the vibe, length and emoji level you choose.
Try it free 🎠A topic into 5, 7 or 10 slides written in reading order, plus the caption to go under them.
Try it free 🎬A Reels, TikTok or Shorts script beat by beat: the line to say and the shot to film beside it.
Try it free 💼One thing you know into three LinkedIn posts that read like a person, not a press release.
Try it free 💡Six post ideas for your account, each with its format, its opening line and why it would work.
Try it freeThe case for generating posts is not that a model writes better captions than you do. It is that the thing standing between most brands and a consistent feed is the blank caption box on a busy day, and a draft you can react to is a different task from a cursor you have to fill.
Most brands do not stop posting because they ran out of strategy. They stop because writing the caption is the small, daily, unglamorous task that loses to everything urgent. A draft you can react to removes that stall, and reacting to three drafts is a different job from starting at a cursor.
Nearly every brand underuses the ideas it already has. The change you shipped, the question you answered twice this week, the thing you learned from a customer call: each is several posts, on several platforms, with different angles. Turning one input into three drafts is the cheapest way to raise output without raising the number of ideas you need.
Posting rhythm compounds, and a gap costs more than a mediocre post does. Anything that shortens the distance between having an idea and having something publishable is worth more than another tool that tells you what you should be posting.
Drafting is the task AI is genuinely good at, and using it for that is no longer unusual. The open question is not whether to draft with a model, it is whether the draft is grounded in anything real about your brand.
A vague idea in, a vague post out. Every time, no exceptions.
The generator cannot know that you moved the roastery. Type the thing you would tell a colleague and the drafts change completely.
Drafting with a model is now ordinary work: SurveyMonkey found 88% of marketers using AI in their day-to-day work. Generated posts still fail for one reason above all others: they are written from nothing. A model given "post about our product" has only the average of everything it has read to draw on, which is exactly what the recognisable AI caption sounds like. A model given the sentence a customer used when they complained, the feature name nobody outside the company knows, or the reason you changed the packaging, writes something only your brand could have posted. The input is the whole game, which is why the notes field on this page matters more than the tone field.
Sarah Levinger, Founder of Tether Insights, names the gap that leaves in most content stacks: “I wish more brands would use the data they have and start analyzing for emotional insights instead of just transactions.” The same customer language that makes a post land makes an ad land, which is why the free hook generator and the AI content brief generator work from reviews the same way. This tool is the drafting step of that loop, run once, on one idea you typed in. If the post you are writing is a photo caption, the Instagram caption generator asks for the scene instead of the idea, and the LinkedIn post generator is this same tool with the platform already set.
A prompt like "our new product" gives the generator nothing to hold. "We cut delivery from nine days to two because we moved the roastery" gives it a post. The idea field is not a keyword box, it is the thing you would tell a colleague, and the more of that you type the less the output reads like it came from a template.
Reach, comments, clicks and sales pull a caption in four directions. Pick the one that actually matters for this post and let the other three go. If you want the same idea working two ways, run it twice with two goals and compare, which is faster than editing a caption that was trying to do everything.
The notes field is where a generated post stops sounding generated. Drop in a review, a support message, the exact words someone used when they complained. Those phrases come back inside the captions, and a sentence a customer actually said outperforms any phrasing a model would reach for on its own.
The useful output is often not the wording, it is seeing that the same idea has three entry points and one of them is better than the one you had in mind. Take the angle, rewrite the words in your own voice, and post that. Regenerating for a fifth version of a weak angle is the slowest way to use this.
The founder writes the posts between everything else. Three drafts from one idea is the difference between posting this week and meaning to.
No back catalogue, no voice document, no rhythm yet. Starting from a draft that already fits the platform beats starting from an empty box and a vague plan.
The same idea has to land on four surfaces in four shapes. Switching the platform and running it again is faster than rewriting a caption four times.
A client hands over one update and expects a week of content. Three angles per idea makes that a review job instead of a writing job.
Every draft returns in the same four parts, so three posts are comparable rather than three loose paragraphs. This is what the tool above produces, not a wish list.
Every draft says how it comes at the idea, so you can see why the three differ rather than reading three variants and guessing.
Opening line, paragraph shape and length follow the surface you picked, because they are not interchangeable.
A specific handful where tags help, and none where they do not. Tag walls date a post and nobody reads them.
No statistic, price, discount or review quote appears unless you supplied it. A generator that makes up numbers hands you something you cannot publish.
Yes. Three drafts per run, no account, no email, no credit card. There is a small cap per page load so the shared model key stays available for everyone, and reloading the page gives you another batch.
Instagram, TikTok, LinkedIn, Facebook and X. The platform is not a label on the same caption: it changes the opening line, the paragraph shape, the length and the hashtag convention, because those differ per surface.
That depends almost entirely on what you type in. The generator is told not to reach for the usual openers and not to stack rhetorical questions, but the thing that makes a post specific is the detail only you have: the real objection, the actual feature, the words your customers use. Paste those into the notes field and the difference is obvious.
No, and that is deliberate. It only states a number, a price, a date, a customer name or a quote if it appears in what you gave it. Generated proof is the most dangerous thing a caption can carry, because you are the one publishing it under your own brand.
No. This tool writes drafts and you copy them. Selzee is not a scheduler, it is the content team that decides what to make and makes it, so the drafts are the output rather than the calendar.
A general chatbot writes whatever the prompt asked for. This one is built around one job: it knows what each platform rewards, it asks for the goal because the goal changes the caption, it returns three genuinely different angles rather than three rewrites, and it refuses to invent proof. That is what the prompt is doing for you.
This is one run from one idea you typed. The product does the same job continuously from your own brand: research, concepts, the content itself, then what it learns from how the content performed. The free tool is the drafting step on its own.
This tool is one run from one idea you typed, and that is useful for getting unstuck. The harder problem is the one underneath it: knowing what is worth posting about this week, which of your customers' words are worth building on, and what the last month of posting actually taught you. Selzee works from your own brand rather than a pasted prompt, so research, concepts, the content itself and what it learns from performance run continuously instead of one caption at a time.
The generator is free right now, with no signup and no credit card. When you want posts that come from your own brand and keep coming, start free.
Start free and let Selzee post from your own brandPrefer to explore on your own? Get started or browse free ecommerce tools.
Keep exploring: the free hook generator, the AI content brief generator, the Instagram caption generator, the LinkedIn post generator.
