For founders
Submission guidelines
Five minutes here will save you a week of back-and-forth. Everything below is what a reviewer actually checks, in the order they check it.
What we accept
AI tools that a reader can sign up for and use today, in one of 9 categories. The categories are closed on purpose: each one implies a ranking, a hub page and a set of judgements we have to keep current, so adding a tenth is an editorial commitment rather than a dropdown entry.
Text to speech · Voice cloning · Music generation · Transcription · Audio editing · Dubbing & translation
General assistant · Reasoning & analysis · Search-grounded chat · Character & roleplay · Open-weights model · Enterprise assistant
AI code editor · Coding agent · Code completion · Code review · Testing & QA · App builder · DevOps & infrastructure
Graphic design · Logo & brand · UI & product design · Presentations · Mockups & 3D · Interior & architecture
Text to image · Image editing · Upscaling & restoration · Background removal · Product photography · Avatars & headshots
Notes & knowledge · Meetings & recording · Automation & workflows · Email & inbox · Scheduling · Project management · Spreadsheets & data
Answer engine · Academic papers · Document analysis · Market & competitor research · Data extraction · Fact checking
Text to video · Video editing · Avatars & presenters · Captions & subtitles · Clipping & repurposing · Motion & animation
Long-form drafting · Copywriting & marketing · Grammar & editing · Paraphrasing · SEO content · Scriptwriting
Not in the list?
Submit anyway and say so in the last box. A category we do not have is useful information — it is how we decide what to add next — and you will get a straight answer rather than a rejection that reads as a judgement of your product.
The quality bar
There is one test, and everything else follows from it: could a reviewer sign up and get value from this within twenty minutes? Not "is it impressive" and not "is it finished" — could a real person with a real task get somewhere. Concretely, that means:
- The URL loads for a signed-out visitor and does not require an invite code.
- There is a way to try it, or a price list that says plainly what happens if you pay.
- It does something a reader could not get free from ChatGPT with a decent prompt.
- Somebody is maintaining it. A changelog, docs or a recent release is enough to show that.
Descriptions
The two description fields do different jobs, and the most common reason a submission goes back is that they have been given the same answer twice. Limits are enforced exactly as stated:
| Field | Length | What good looks like |
|---|---|---|
| Tool name | 2–60 | Cline — not "Cline AI Coding Agent (Free!)" |
| Official website | up to 300 | https://cline.bot |
| One-line description | 60–180 | "Turns a long PDF into a searchable set of notes with citations back to the page." — concrete, checkable, no adjectives. |
| Full description | 300–2500 | Name the real constraint — a rate limit, a missing integration, a workflow it cannot do yet. We will find it anyway, and finding it ourselves is worse for you. |
| Why should PromptHive feature this tool? | 100–1200 | "Cursor and Copilot both charge a flat fee that hides model cost. We pass it through, so a light user pays $8 instead of $20." — a specific, testable claim. |
| Plans and prices | 40–1200 | State the cap, not the marketing word. "Unlimited*" with an asterisk is the single most common correction we have to make. |
| Logo URL | up to 300 | https://acme-copilot.example/brand/mark-512.png |
| Company name | 2–100 | The legal entity behind the tool, or your own name if you are an individual. |
The single most useful thing you can write
Name your product's weakness in the full description. Every review we publish has a "watch-outs" section, including the tools we rate highest, so the question is never whether the limitation appears — only whether we found it ourselves after you implied there wasn't one. Founders who state theirs get reviewed faster and rated more generously, because we can tell the rest of the submission is honest.
Logo requirements
The logo is rendered in a square box, and at 32 pixels on the directory page. A horizontal lockup with the company name beside the mark becomes an illegible smear at that size — send the mark on its own.
Send a link, not a file. We ask for a direct URL to the image — the address that ends in.png or .svg, not the page it appears on. Right-click the image and copy the image address. Anything we publish is re-hosted by us, so the link only has to work while your submission is being reviewed.
- Format: PNG, SVG, WEBP, JPEG
- Size: at least 128px on the shorter side, up to 2048px
- Shape: square, or close to it — we reject anything more than 1.25× wider than tall
- Background: transparent. A white square looks like a bug on our paper canvas.
SVG is preferred — it is the best source for the raster sizes we generate. Export a flattened file: an SVG carrying scripts or external references is refused, and we check before anything is published.
Screenshot requirements
- How many: up to 4
- Format: PNG, WEBP, JPEG
- Size: at least 640px wide
- How to send them: one direct image URL per line
Show the product working on something real. A signed-in view with actual content in it is worth four hero images of your landing page, because the reviewer is trying to answer "what is it like to use this" and a marketing site cannot tell them.
Reasons a submission is turned down
These are the actual reasons, in roughly the order they occur. None of them is a judgement of you, and all of them come with an email that says which one applied.
It is not a product yet
A waitlist, a landing page, a Notion doc or a private beta we cannot reach. We review things readers can use today. Come back when they can.
The pricing is hidden
"Contact us for pricing" as the only option. Every review carries verified prices with the date we checked them, and we cannot verify a number that does not exist publicly. Enterprise-only tiers are fine — an entirely private price list is not.
It is a thin wrapper
A prompt in a box in front of somebody else's API, adding no workflow, no data, no interface worth having. There is nothing wrong with building on a model — most good tools do — but the value has to be yours.
The description is written for a crawler
"Revolutionary AI-powered platform leveraging cutting-edge technology" tells a reader nothing and tells us you have not thought about who they are. We will send it back once; twice and we stop.
The claims do not survive checking
A stated limit that is not the real limit, a free tier that needs a card, a model list that includes something it does not use. A submission that misstates its own facts is rejected rather than corrected — if the easily checkable parts are wrong, we cannot stand behind the rest.
It duplicates something we already cover, with no reason to prefer it
The catalogue is capped deliberately. A tool that does what a listed tool does, slightly differently, needs an answer to "why would a reader pick this one?" — and that answer has to be more specific than "better UX".
The category does not exist here
We publish in 9 categories and add new ones only as a deliberate editorial decision, because each one implies a ranking we have to maintain. If yours is not on the list, tell us and we will say honestly whether it is coming.
How long it takes
- First look
- Within five working days, usually sooner. This is a triage pass: does it load, is it in a category we cover, is the submission complete.
- Full review
- One to three weeks after that. Somebody uses the tool, reads the docs and verifies every price. This is the part that cannot be hurried, and the reason a listing here carries weight.
- If we need something from you
- The clock stops until you reply. We hold a submission open for 30 days after asking, then archive it — you can always submit again.
- Publication
- After approval, when the review 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.
What we will never do
Worth stating explicitly, because every one of these is standard practice elsewhere in this category:
- Charge for a listing, a review, a rating or a position. Ever.
- Offer a paid fast-track, "featured" slot, or bid-per-click placement.
- Let a submission influence a rating, a ranking or a comparison verdict.
- Sell, share or rent what you send us.
- Remove a watch-out because a vendor asked. Correct one that is factually wrong — always.
The reasoning is in how we review AI tools, which is the document to read if you want to know how a rating gets decided rather than how to submit.