Zway.ai

What we run

Product launches growth, run for you.

A launch is won in the four weeks before it, not on the day. The list of people who will actually care has to be built by hand, one conversation at a time. Then somebody has to sit on the launch for fourteen hours answering every comment live, in public. Zway does both parts.

Last reviewed 27 August 2026

The failure mode

What breaks when AI does this alone.

A launch has one shot and a fixed window, which is the worst possible combination for automation. The preparation is entirely relationships, and an automated message to somebody who has never heard of you is worse than no message, because it spends the introduction. On the day, the comment thread moves faster than any review cycle. A generic reply to a sharp question does more damage than silence, since everybody reading can tell, and the thread stays up permanently as the first result people find.

What our operators do here

  • Build the launch list by hand over the weeks before, through real conversations
  • Write the tagline, the assets, the first comment, and the maker story in your voice
  • Line up the people who genuinely want to see it, without asking anybody to vote
  • Run the day live, answering every comment within minutes for the full window
  • Handle the hostile comments and the feature requests in the same steady tone
  • Sequence the surfaces so the launch does not collide with itself
  • Write the honest post mortem, including the parts that did not work

Volume

What ships in a typical week.

Typical weekly output on the Product launches channel
WhatTypical volume
Pre-launch conversations started by hand20 to 40 per week
Launch assets drafted and revisedFull set, three weeks out
Comments answered on launch dayEvery one, live
Follow-up surfaces sequenced after the day2 to 4

Volumes are typical rather than contractual. They move with what the channel is actually returning.

What good looks like

The metric that matters

Comments from people who clearly understood the product, and signups that survive week two

Realistic timeline

Four to six weeks of preparation for one launch day. The day itself is the smallest part.

The launch is decided before launch day

Most founders treat a launch as an event. It is the tail end of four to six weeks of work, and almost everything that determines the outcome has already happened by the time the page goes live.

What happens on the day is that a group of people either shows up or does not. Whether they show up depends on whether anyone told them, personally, in the weeks before, in a way that made them want to. That is a list built one conversation at a time, and it cannot be assembled the night before.

The second half is presence. Someone has to be in the comment thread for the entire window, answering everything, in public, with no ability to edit their way out of a bad reply.

What the four weeks actually contain

WhenWhat happensWho does it
Four weeks outConversations started with people who would genuinely careOperator
Three weeks outTagline, gallery, first comment, and maker story draftedAI layer, operator rewrites
Two weeks outObjection list built from your real support and sales questionsAI layer
One week outPeople told the date, personally, with no ask attachedOperator
Launch dayEvery comment answered live, for the full windowOperator, with you
Week afterFollow-up surfaces sequenced, post mortem writtenBoth

The row that looks smallest is the one that takes the most hours. Fourteen hours of live thread is a full working day of writing under time pressure, and it is why launches run by busy founders alone tend to go quiet by lunchtime.

Launch day is a live conversation, not a broadcast

The comments arrive faster than any approval process. Someone asks how you compare to a competitor, someone else says they built this in a weekend, someone reports a bug in the signup flow while a hundred people are in it.

Each of those needs an answer within minutes, and each answer is permanent and public. The bug gets acknowledged plainly. The competitor question gets an honest comparison including where the competitor is better, because everyone reading can tell when it does not. The dismissive comment gets a short, genuinely friendly reply, because arguing in a launch thread is how a good day becomes a screenshot.

This is the part that no draft-and-schedule product reaches, and it is not close. The window is fixed, the audience is watching, and there is no revision cycle.

More than one surface, in the right order

Product Hunt is the surface people mean, but it is rarely the only one, and the others have their own rules.

Hacker News is the sharpest example. Its guidelines exclude blog posts, sign-up pages, newsletters, and landing pages from Show HN, so a launch page is often not eligible in the first place. The FAQ is explicit that asking for upvotes gets submissions and accounts penalised or banned. Those constraints change the sequencing, not just the copy, and we cover them properly on the Hacker News page.

Relevant subreddits, community Slacks and Discords, your own list, and your own audience all have different tolerances for the same announcement. Firing all of them at once looks coordinated, which is exactly the impression you do not want. An operator spaces them and adapts each one.

What we will not do

We will not ask anyone to upvote, run accounts we do not disclose, or coordinate a group of people to arrive at the same minute. Every launch surface treats that as manipulation, and getting caught costs you the account and the launch at the same time.

We will also tell you when the product is not ready, which is the least popular thing on this page and the most useful.

Product launches questions

When should we launch on Product Hunt?
When the product can survive strangers using it unsupervised, and not before. A launch is a one-time introduction to an audience that will not come back for a second attempt, so launching into a broken onboarding flow spends the only shot you get. If the product is ready, we want four to six weeks of preparation ahead of the date.
Can you guarantee a top spot or a specific ranking?
No, and anybody who does is either buying votes or guessing. Ranking depends on who else launches that day, which nobody controls. What we can control is that the right people know it is happening, that the assets are good, and that every comment gets a real answer. Those are the parts that actually convert attention into users.
Can an AI run a launch?
It can do the preparation that is research and writing. Finding the right people, drafting the assets, preparing answers to the questions you will predictably be asked. It cannot build the relationships in the weeks before, because those come from actual conversations, and it cannot sit in a live thread for fourteen hours making judgment calls in public with no undo button.
Do you ask people to upvote?
No. On Hacker News the FAQ is explicit that soliciting upvotes gets submissions, accounts, and sites penalised or banned, and we treat that as the standard everywhere rather than a rule to route around. What we do instead is make sure the people who would genuinely want to see the launch know the date. Whether they vote is up to them.
What happens if the launch flops?
It happens, and it is recoverable, because a launch is one surface rather than a strategy. We write up what actually went wrong, which is usually the positioning rather than the tactics, and the comment thread is often the most useful feedback the company has ever received. Then the same work feeds the channels that compound instead.

See what we would run on Product launches.

Thirty minutes. You leave with the thirty-day plan either way.

Book a demo

30 minutes. You leave with the plan either way.