How to Get Your AI Tool Featured on PromptHive

We read every submission that comes through /submit/. A reviewer opens the product, uses it, checks the links and verifies the prices. Some submissions make that straightforward and some make it slow, and the difference is almost entirely down to four or five fields.

This is what a reviewer is actually looking at, in the order they look at it.

Before you open the form

Three things decide the outcome before you type anything, and none of them are fixable in the form.

Can a stranger use it today? Not a waitlist, not an invite code, not a demo video. A reader has to be able to sign up and get somewhere. This is the single most common reason a submission is turned down, and it is not a judgement of the product — it is that we review products rather than roadmaps. If you are two weeks from launch, submit in two weeks.

Is there a public price? Every price we publish is read off the vendor’s own pricing page and stamped with the date. “Contact us for pricing” as the only option means there is nothing to verify, so there is nothing we can publish. An enterprise tier priced on application is completely normal; an entirely private price list is a blocker.

Does it do something the tools we already cover cannot? We cap the catalogue at around 150 on purpose, so every addition displaces attention from something else. This is the bar that catches good products: a competent tool that does what ChatGPT or Notion AI already does, slightly differently, needs a better answer than “better UX”.

The three fields that decide everything

The form has around thirty fields. Three of them are the review.

The one-line description

60 to 180 characters, and the temptation is to spend them on adjectives. Don’t. Spend them on the mechanism.

✗ “The revolutionary AI-powered platform transforming how modern teams collaborate on content at scale.”

✓ “Turns a long PDF into a searchable set of notes, with every claim linked back to the page it came from.”

The second one tells a reviewer what to try in the next five minutes. The first one tells them nothing, and — more expensively — tells them the rest of the submission is probably written the same way.

A useful test: delete every adjective and see whether the sentence still says what the product does. If nothing survives, that was the problem.

The full description

300 to 2,500 characters. Four paragraphs is plenty. Cover, in this order:

1. The problem, stated as a situation rather than a market. “Support teams answer the same forty questions in slightly different words” is a problem. “The customer experience space is being disrupted” is not.
2. How it actually works. Which models, what it does with your data, where the work happens. Readers increasingly ask this and almost nobody publishes it.
3. What it is genuinely good at. Specifics. One thing it does better than anything else beats five things it does adequately.
4. Where it falls down. See below — this is the paragraph that matters most.

Why should PromptHive feature this tool?

This is the field founders tell us was the hardest, and it is the one we read most carefully, because it is the only place you get to make the argument yourself.

What works: a specific, checkable claim about a comparison. Something like “Cursor and Copilot both charge a flat monthly fee that absorbs the model cost. We pass it through, so a light user pays $8 instead of $20 — and can see what each task cost.” That is testable in twenty minutes, it names the comparison, and it gives the reviewer the axis to judge you on. It is roughly the argument that got Cline reviewed.

What does not work: “we are the leading platform for…”, a list of logos, a funding announcement, or your traffic numbers. None of them answer the question a reader has, which is why would I pick this one.

Name your own weakness

The most counter-intuitive advice we can give, and the most reliable.

Every review on PromptHive carries a watch-outs section. All of them — including the tools we rate highest, because a page with no watch-outs is a page nobody believes. So the limitation is going on the page regardless. The only variable is whether we found it ourselves after your submission implied there wasn’t one.

Submissions that name the constraint get through faster and, in our experience, get rated more generously. Not out of sympathy — because everything else in the submission becomes credible at once. A founder who tells us their tool cannot handle multi-page documents yet has just told us they know their product and are not managing us.

The pattern to avoid is the one that reads as confident and lands as evasive: “Some users initially find the interface unfamiliar.” That is not a weakness, that is a compliment in costume. Try: “There is no mobile app, and the export only does CSV. Both are on the roadmap and neither is done.”

Assets

Two rules, both learned from fixing them by hand.

The logo is the mark, on its own. It renders in a square box, at 32 pixels wide on the directory page. A horizontal lockup with your company name beside the mark becomes a smear at that size. Square, at least 128px, transparent background, SVG if you have it — SVG is the best source for the raster sizes we generate. The form checks the aspect ratio and will tell you before you finish, which is a lot better than an email three days later.

Screenshots should show the product working. A signed-in view with real content in it is worth four of your landing page hero. The reviewer is trying to answer “what is this like to use”, and a marketing site cannot tell them. Up to four, 640px wide or more.

Mistakes we see repeatedly

• The same text in both description fields. It means one of them was not answered, and it is immediately obvious.
• Prices that do not match the pricing page. We check. A mismatch turns a review into a round of email, and if the easily checkable facts are wrong we have to assume the rest might be.
• “Unlimited” with an asterisk somewhere else. State the cap. The shape of a cap is fine — “throttles after a burst, then drops to a smaller model” is an honest and useful answer. A hidden asterisk is the single most common correction we have to make to a published page.
• A category that does not exist here. We publish in nine, and we add a tenth as a deliberate editorial decision rather than a dropdown entry. If yours is not on the list, say so in the last box — it is genuinely useful information and you will get a straight answer instead of a wait.
• Ticking every access checkbox. “Free plan, free trial, no card, API, open source, self-hosted” is rarely all true at once, and a contradiction the form catches (a free plan on a Paid pricing model, for instance) is the first thing a reviewer notices.
• A free-mail address with no other verification. Perfectly fine for a solo founder, and it does not count against you — but an address at your own domain is the quickest way for us to confirm you are connected to the product, and it saves a day.
• Submitting a tool we already list. Check first. If it is already there and something is out of date, submit anyway and say what — we route it to the editorial team as a correction, which is usually faster than a fresh review.

What happens after you send

A first look within about five working days: does it load, is it in a category we cover, is the submission complete. Then the review itself, one to three weeks, which is somebody using the tool.

You will hear from us at every stage, including if the answer is no. A rejection names the specific reason — and if that reason changes, submit again.

If we approve it, the review gets written and published when it is written; we do not publish to a schedule, so we will not give you a date we might miss. You get the link the day it goes live, and from then on you have a correction channel: tell us a fact is wrong and it gets fixed that week, with the check date moved.

The short version

Everything above compresses to five sentences.

Make sure a stranger can use the product and see a price. Describe the mechanism, not the market. Name the competitor you expect to be compared against, and the axis you win on. Name your own weakness before we find it. Send a square logo and a screenshot of the thing actually working.

Then read the submission guidelines — five minutes, and it is the same list from the reviewer’s side — and submit your tool.