Chrome extension · launching soon

Reply in your voice.
Not ChatGPT's.

ReplySathi drafts three replies for any tweet — in your words, your rhythm, your slang. Hindi-English included. You pick one, tweak it, hit reply.

Free plan forever, no card. Built for indie hackers, founders and creators growing on X the honest way — by replying.

Thirty seconds of setup. Then it's just a button.

  1. 01

    Sign in, teach it your voice

    Sign in with Google — no API key, no card — then paste three to five of your own tweets into the extension. That's the entire setup. It reads your sentence length, your punctuation habits, whether you capitalise.

  2. 02

    Hit ✦ on any tweet

    A small spark sits in the tweet's action bar, next to reply and like. Click it and three drafts appear right under the tweet. Works on the timeline, on profiles, inside threads.

  3. 03

    Pick one and send it

    Click a draft — it lands in X's own reply box. Change a word, change all of them, or write something else entirely. You press Reply. Always.

Every other AI reply tool has one voice.
It's this one.

You've seen it a thousand times in your replies. Everyone can spot it. It's why AI replies get ignored — and why they quietly cost you followers.

Generic AI reply tool
Great point! Execution truly is everything. Planning without action leads nowhere. 🚀 Keep shipping! #startup #founders #buildinpublic

Four tells: the opener, the emoji, the hashtags, the zero information.

ReplySathi, trained on your tweets
guilty. sat on a landing page for 3 weeks polishing copy. launched it, got the same feedback in a day.

Lowercase because you write lowercase. Specific because you're specific.

Hinglish mode, on the same tweet
bhai yeh sach mein hota hai. 3 hafte landing page pe gaye, launch kiya toh ek din mein wahi feedback mil gaya.

Roman script, natural code-switching, only where it actually fits the tweet. No forced Hindi. No competitor ships this.

Competitors start at $39/mo.
We start at free.

Free

$0forever

To find out whether it sounds like you.

  • 6 replies a day, 60 a month
  • Voice profile
  • All five tones
  • Hinglish mode
Start free
Most people land here

Starter

$9/month

For a daily reply habit that actually compounds.

  • 30 replies a day, 400 a month
  • Everything in Free
  • No API key, ever
  • Cancel any time
Join the waitlist

Pro

$19/month

For people who reply like it's a job. Because it is.

  • 60 replies a day, 700 a month
  • Written by our best model
  • New features before anyone else
  • Your message reaches the maker
Join the waitlist

Prices in USD; pay in rupees if you're in India. Annual billing is two months free. Cancel any time. Paid plans open on launch day — until then everyone starts on Free, and the waitlist locks founding-member pricing.

Straight answers

Is this against X's rules?

ReplySathi doesn't automate posting. It drafts text, puts it in X's own reply box, and stops — you read it, edit it, and press Reply yourself. That's the same category as a keyboard suggestion, not a bot. X's automation rules target unattended posting, spam and bulk engagement, and we deliberately make all three impossible. We'd still say read X's rules yourself and use it like a person, not a machine gun.

Does it auto-post?

No. There is no code path in the extension that clicks X's Reply button. That is a deliberate design decision, not an unshipped feature, and it isn't coming.

What data do you store?

Very little, and all of it tied to the Google account you sign in with: your email and name from that account, which plan you're on, and how many replies you've generated today. The daily counter deletes itself after 48 hours. There's also a plain signup log — an email and a date per new account — and that is the whole database. We don't store the posts you click or the replies we draft, we never see your DMs, and we can't post as you — there's no X login, no X API token, nothing attached to your X account.

What happens to the posts I click ✦ on?

The text of that post — plus a bit of context like the author's name, quoted post or thread above it, and your voice profile — goes to our server, which builds the prompt and sends it to the model provider (Anthropic today). The drafts come back to your browser. Nothing on that round trip gets written to a database: not the post, not the drafts, not your voice profile, which lives in your browser's extension storage and is sent with each request rather than kept by us. The provider holds API traffic briefly for abuse monitoring under its own policy; nothing there trains a model, and nothing on our side is sold to anyone.

Will the replies actually sound like me?

In proportion to the effort you put into the voice profile. "Casual and friendly" gets you a generic bot. Four of your real tweets pasted in gets you something your followers won't clock as AI. It takes two minutes and it's the whole product — there's a worked example of a good profile on the getting-started page.

Which browsers work?

Chrome, Edge, Brave, Arc — anything Chromium-based. Firefox and Safari aren't supported yet.

Do I need an API key?

No, and there's nowhere to put one. Earlier builds asked you to paste your own key; that's gone. You install the extension, sign in with Google, and click ✦ — we run the model on our own key and our own servers. The free plan is 6 replies a day, 60 a month, with no card, and there's no OpenAI console, no billing dashboard, no usage bill to be surprised by.

How do I sign in?

From the extension itself. Click ✦ on any post, or open the extension popup, and hit Sign in with Google. It opens Google's own window, you pick your account, it closes. No password to make, no confirmation email to hunt for. There's nothing to sign into on this website — the account only exists inside the extension.

Get in before launch

Early access, founding-member pricing locked for a year, and a say in what gets built next. No spam — you'll hear from us when there's something to open.