Launches fail quietly. Not with a crash, but with a post that nobody sees, a page that does not say what the thing is, and a founder refreshing analytics at midnight. Almost all of it is preventable with preparation, and AI makes the preparation affordable for one person. This is the plan: what to have ready, where to show up, what to say, and what to do when the first day is slower than you hoped, which it usually is.
Free to start. Launch copy is where the latest models sound like you instead of like marketing.
The most common launch mistake is launching a product that has not been used by a stranger. Before any of the below, five people you did not recruit personally should have used it, and at least two should have paid or committed. If that has not happened, go back to validation; a launch will only make the silence louder.
The second most common mistake is launching once. A launch is not a day; it is a sequence of moments where you show the product to a new group of people. Product Hunt is one moment. A post in a community is another. An email to the list is another. Plan four or five of them, spread over weeks, and each one gets better because the last one taught you something.
Everything below lives in the launch Project alongside the product's own Project context, so every asset is written from the same customer, the same offer, the same voice.
One sentence: [Product] helps [who] do [what] without [the annoying thing]. Every asset comes from it. Then have the AI list every asset the launch needs, in the order a customer meets them: name, landing page, pricing page, screenshots, privacy policy, sign-up flow, onboarding emails, help page, support address. Tick them off in the Project.
Headline, one sentence, a real screenshot, three benefits, one button. Drafted from your customer interviews if you have them, in their words. Have a second model review it with one question: "What here would make a sceptical [customer] not trust us?" Then take the five screenshots the AI says would convince someone, from the real product.
Ask for the twenty places your customers already gather and what kind of post is welcome in each. Choose five. Spend this week being useful there without mentioning the product. Meanwhile draft, in your voice: a community post that leads with the problem, a launch-site description, a personal message to the people who tried the beta, and the one-line answer to "what are you working on?".
Sign-up works on a phone. Payment works with a real card. The confirmation email arrives and does not land in spam. Error reporting is on. Analytics is on. Someone who has never seen the product can go from the landing page to the first win in ten minutes; watch them do it. Ask the AI for the full pre-launch checklist for your platform and go through it twice.
Post in the five places, spaced through the day. Reply to every comment within an hour; the replies are the launch. Keep a file of every question, objection and bug that comes in. That file is week one's work.
| Channel | Good for | What to prepare | Honest expectation |
|---|---|---|---|
| Communities where your customer already is | Almost every first launch. Subreddits, Slack and Discord groups, industry forums, Facebook groups. | A week of being useful first; a post that leads with the problem, not the product. | Your first ten to fifty users. The best channel by far for niche products. |
| Your beta users and interviewees | Products that came out of validation. | A personal message to each, asking for one specific thing: a share, a review, a referral. | Small numbers, high quality. These are the people who refer others. |
| Product Hunt and similar launch sites | Tools for people who work in tech or startups. | A clear tagline, five screenshots, a first comment that tells the story, friends ready to comment early. | A spike of curious visitors, few of whom are your customer. Useful for feedback and a badge, rarely for revenue. |
| A personal post on LinkedIn or X | Anything; it costs nothing. | The story of why you built it, with one screenshot. Not a feature list. | Depends on your network. Reposts from people in the industry matter more than likes. |
| Email to a list you built during validation | Products that ran a landing-page test. | Three short emails: it is live, what people are saying, last chance for the launch price. | Your best conversion rate of any channel, if the list is real. |
| Paid ads on day one | Almost nothing. | Do not. You do not yet know which message works. | Money spent learning what the free channels would have taught you. |
| Press | Products with a genuinely new angle or a local story. | A short, specific pitch to five named journalists who cover your niche. | Usually silence; occasionally a piece that outperforms everything else. Low cost to try. |
It usually is. The launches that look like overnight successes were mostly four or five quiet launches, each a little better. Here is the routine for the first week.
Paste in visits, sign-ups, first-win completions and payments per channel. Ask what the funnel says. Many visits and no sign-ups means the page is unclear; sign-ups with no first win means onboarding is broken. Each has a different fix.
Take the launch-day file of questions and bugs. Have the AI group it and rank it. Fix the top item with the product's own Project, ship it, and tell the people who raised it. That reply is often the moment they become a customer.
Pick the next community on the list. Rewrite the post using what you learned: the question everyone asked, the objection that came up, the quote from a user who liked it. Every launch after the first has evidence in it.
Not visits or sign-ups. People who came back on day three. If nobody returns, no amount of launching helps, and the answer is in the product. If people return, every launch compounds. Ask the AI for the simplest way to measure that number for your product, and put it on a page you look at every morning.
Readers can tell generated marketing at a glance, and they scroll past it. The latest models, given three paragraphs you wrote yourself and the actual words your customers used in interviews, write launch posts and emails that sound like you. Older models write like a press release. For a launch, where every sentence is read by a stranger deciding whether to trust you, that difference shows up directly in sign-ups.
Newer models hold your whole Project in view instead of the last few messages, reason through trade-offs instead of picking the first plausible answer, and are wrong less often and with less confidence. For the questions on this page, that is the difference between advice that sounds right and advice you can act on.
Pro from $20 a month, cancel any time. The free plan stays free.
Add the positioning sentence, three paragraphs in your own voice, and any customer quotes. Then start here.
When five strangers have used it and at least two have paid or committed, and a new person can get to the first useful result in ten minutes without help. Not when the feature list feels complete; it never will.
If your customers are people who work in tech, yes, as one of several launch moments. If they are plumbers, restaurant owners or clinics, the community where they already are is worth ten Product Hunts.
Close to nothing. The channels that work for a first launch are free and require time, not money. Paid ads belong after you know which message converts, and the launch is how you find that out.
Read the funnel. Visits with no sign-ups is a page problem; sign-ups with no activity is an onboarding problem; no visits is a channel problem. Each has a fix, and the second launch a week later usually goes better than the first. See the week-after section above.
It can draft them, and with your own writing and your customers' words in the Project the drafts will sound like you. Cut anything that sounds like marketing before you post. You are the one people are deciding to trust.
Your first Project, file uploads and the major model families. The latest models, more messages, generated documents and images are on paid plans. See plans.
Start the launch Project today: positioning sentence, assets list, five channels. The product's own Project keeps everything consistent. Free to start.
Also: start a business with AI · validate a business idea with AI · how to build an app with AI
Continue learning
Run the same prompt through ChatGPT, Claude, Gemini and Grok before trusting one answer.
Core featureLet several models draft, challenge and verify each other instead of trusting one answer.
ProjectsKeep files, instructions, code and chats together so every model works from the same context.
FeaturesModels, Projects, collaboration, verification, Studios and images in one workspace.
HubIs the idea good, who else does it, what must it do, what will it cost: answered, then built in a Project.
ValidationA five-rung ladder with pass marks, from scoring to pre-selling.
Market researchSourced market size, segments, pricing and competitors with Perplexity Sonar Pro.
PlanTen sections, one page each, bottom-up financials, kept alive in a Project.
Online businessA real problem, a small product or service, a demand test, honest costs.
IdeasSixteen narrow, startable ideas with who pays and what you build.
StartupsTwelve vertical ideas with a wedge and a why-now.
FoundersOne Project for research, code, decisions and launch work.
