AI reply generator for X
Click the PostInk mark in any reply box on X. It reads the post and the thread, writes three replies in your voice, sets aside any that sound like a bot, and puts the best one in the box. Then you press Reply.
Each card pairs the reply PostInk showed with a draft its check blocked for the same post. The first three are real replies to real posts.
Grok has bots OpenAI has dots Anthropic has ants
This is a game changer for how we think about AI branding. Who else has noticed this?
PostInkgoogle has googles
The post you’re answering, the posts above it, and the text inside any screenshot. All of it is read on your device.
Three different angles in a single call, in the voice it learned from your own posts.
Every draft is checked for the phrasing that gives generated text away. Any draft that has it is blocked.
The most human draft that passed goes into X’s reply box. Edit it, ask for another, or post it.
People on X can spot a generated reply in a second. So can you.
Most AI reply generators open with praise (“Great point!”), restate the post, add a tidy lesson, and close on a question nobody asked. Each of those is a pattern. One reply like that is harmless. Ten a day, and your account starts to read like a bot, which is worse than not replying at all.
PostInk treats those patterns as failures. A draft that opens with praise, says “it’s not X, it’s Y”, reaches for words like delve or game changer, or asks for engagement is discarded before you see it. If none of the three drafts pass, it writes one more. If that one fails too, it tells you why instead of handing you a bad reply.
You can run the same list of tells on any text with the free AI writing checker.
The reply side panel drafts replies to the posts you actually stop on, so you can answer five people in the time it takes to answer one.
Open the panel and scroll X as you normally would. The posts you pause on get a draft waiting beside them. Approve one, it goes into that post’s reply box, and you press Reply. A daily cap and a minimum gap between replies are built in. You can lower them, but you can’t remove them.
Looking for auto reply on Twitter?
Tools that reply for you, unattended, are what X’s automation rules exist to stop, and accounts that use them tend to get limited. PostInk is the safe version. It does the writing, and the posting stays with you.
| Typical AI reply generator | PostInk | |
|---|---|---|
| Where you use it | A website you paste the post into | Inside X’s own reply box |
| What it reads | The text you paste | The thread, plus text in screenshots |
| Whose voice | A generic friendly tone | Yours, learned from your own posts |
| Drafts that sound like AI | Shown to you | Blocked before you see them |
| Who presses Reply | You, after copying and pasting | You, with the reply already in place |
| Price | Often a monthly plan | $3.99 once, after 7 days free |
PostInk has no AI of its own and no server between you and the model. Connect OpenRouter in one click (new accounts start on a free model), paste a key from OpenAI, Claude, Gemini or another provider, or run a model on your own computer with Ollama. A reply uses a few hundred tokens, so a day of replies on a strong model usually costs cents. A local model costs nothing.
PostInk is free for 7 days, with every feature and no card. It runs on your own AI key, and connecting OpenRouter starts you on a free model, so the trial costs nothing at all. After the trial, it’s a single $3.99 payment.
No. It writes the reply and places it in X’s reply box. You read it and press Reply yourself. Nothing is posted without you, which is also what keeps your account inside X’s rules on automation.
PostInk learns your voice from the posts and replies you wrote yourself on X, as you scroll past them, and writes in that style. In Voice settings, you can see what it learned, forget any of it, or turn learning off.
Yes. PostInk reads the text in screenshots on your own device, so a reply to a screenshot of a chart or a headline knows what the screenshot says. The image is never uploaded.
The one you choose. Use your own key with OpenRouter, OpenAI, Claude, Gemini, Grok, Mistral, DeepSeek and others, or run a model on your computer with Ollama. Your key stays in your browser and goes only to that provider.
The reply side panel has a daily cap of up to 50 approved replies, with at least 20 seconds between them. You can lower both, but you can’t remove them, so a busy afternoon never looks like a bot.
PostInk drafts in your voice, throws out anything that reads like a bot, and schedules only what you approve. It runs in Chrome with your own AI key or a free local model.
$3.99 once, after 7 days free