How to Get Your First 100 Users for an AI Product

TLDR

• Building is now the easy part. AI let you ship in a weekend. Distribution did not get easier, so it is now the real bottleneck. Plan to spend more time getting users than you spent building.
• Go where your audience already complains. Niche subreddits, Slack groups, and Discords beat your own Twitter followers. One founder got 42 of their first 100 customers from Reddit alone (Indie Hackers).
• Do things that do not scale. Hand-recruit your first users, DM people with the problem, and set them up personally. Paul Graham calls the manual setup a Collison installation (paulgraham.com).
• A launch is a spike, not a strategy. Product Hunt can send hundreds of signups in a day and still produce almost no revenue if your funnel is not ready.
• Pick one channel, run it weekly. Judge it on paying users, not upvotes. Then ask every early user two questions and turn them into referrals.

Building was the easy part. Now comes the hard part

AI tools collapsed the cost of shipping software. You can describe an app and have a working version by the weekend. That is genuinely new. What did not change is the part that was always hard: getting a stranger to find your product, try it, and stay. If anything it got harder, because thousands of other people can now ship a similar thing just as fast.

So the honest starting point is this. Nobody is waiting for your product. There is no path where you post a link once and users pour in. You have to go and get the first hundred by hand, one small channel at a time. The good news is that a hundred is a small, reachable number, and you do not need an audience or a budget to get there. You need consistency and a willingness to talk to people.

Channels ranked for a cold start

From a standing start with zero followers, these are the channels in rough order of how reliably they produce your first paying users. The winner for most solo builders is not the flashy one. It is showing up in the places where your specific user already describes the problem you solved.

Go where your audience already argues about the problem

Your first instinct is to post to your own social following. Resist it. Your followers know you, not the problem you solved. The people you want are in communities organized around the pain: a subreddit for freelancers if you built an invoicing tool, a Discord for a tool your users depend on, a Slack for a specific profession.

The founders of PostKing, an AI content tool, tracked exactly where their first 100 customers came from. Reddit produced 42 of them, their single biggest channel. Two coordinated launches produced another 38. Hours spent submitting to SaaS directories produced three signups and zero paying customers (Indie Hackers). Community and launches did the work. The passive channels did not.

The rule that separates a welcome post from a ban is simple: lead with the problem, not the product. Spend a week or two answering real questions and sharing what you know, with no link. Learn the culture. When you do mention your tool, disclose that you built it. A Reddit go-to-market guide puts it plainly: contribute genuinely, tailor each post to the specific community, stagger posts across three to five subreddits over a week rather than pasting the same thing everywhere, and reply to every single comment with a real answer (reddgrow).

The people you want are in Slack communities for solo agency owners, Discord servers for indie freelancers, and threads where your exact user complains about their exact problem. One niche Slack workspace can beat 5,000 Twitter followers.

Do the things that do not scale

The most reliable way to get your first users is also the least glamorous: recruit them one at a time, by hand. This feels wrong because it obviously will not scale to a million users. That is the point. You do it because it teaches you things automation never could, and you stop once growth carries itself.

The most common unscalable thing founders have to do at the start is to recruit users manually. Nearly all startups have to. You can't wait for users to come to you. You have to go out and get them. — Paul Graham, Do Things That Don't Scale

Graham's favorite example is Stripe. When someone agreed to try it, the Collison brothers would say give me your laptop and set them up on the spot, rather than emailing a signup link and hoping. At YC this is now called a Collison installation (paulgraham.com). Maximum friction for you, zero friction for the user, and every setup shows you exactly where the product confuses people.

Two concrete moves for a solo builder. First, make a warm-30 list this weekend: thirty people who genuinely have the problem, from past colleagues to community members you have seen describe the pain. Message them individually, not as a broadcast. Second, offer concierge onboarding to every early user: a short call where you set up their account and watch them use it. Derrick Reimer did exactly this for SavvyCal, giving personal onboarding to every early user, which both improved the product and turned those users into evangelists (SaaS Opportunities).

This is the same lesson YC teaches: finding your first users is more of a search problem than a persuasion problem, and targeted personal outreach beats any billboard. Gustaf Alstromer, former head of growth at Airbnb, walks through the tactics for both B2B and B2C.

Launch where people look for new tools, but expect a spike, not a business

Product Hunt, Peerlist, and Show HN concentrate people who actively want to try new things. A good launch can deliver a wave of your first users in a single day. The trap is mistaking that spike for traction.

One founder launched BrandingStudio.ai on Product Hunt, hit #6 of the day with 168 upvotes, and pulled in more than 400 signups over four days. Revenue from all of it: $237, from a single paying customer (Indie Hackers). The launch was not a failure. It surfaced 400 real people who showed him exactly where his funnel broke. But it was not, on its own, a business.

Two rules make a launch pay off. First, do the pre-work: spend two to three weeks before launch participating in the community, supporting other launches, and lining up people who will show up on the day. Second, have billing and your first real value moment ready before you launch, not after. Another founder watched session recordings during his launch and found people were reading his page carefully then leaving unconvinced. He rewrote the page mid-launch and got his first trial that evening (Indie Hackers). Launch day is a live usability test. Instrument it and watch.

Build in public and turn early users into word of mouth

Building in public means posting the real journey: what you shipped, what broke, screenshots, revenue, and honest failures. It is slow at the start and it is not a substitute for outreach. Be honest with yourself about the timeline. Farid Shukurov posted daily on X for six straight months and gained about 200 followers, roughly 33 a month, before it started to compound into a micro-SaaS portfolio that has done six figures in lifetime revenue (MRR Story).

What made building in public convert for him was pairing it with the unscalable stuff: every time someone started a free trial, he sent a personal DM, thanked them, and asked about their goal. The public posts built trust at a distance. The DMs closed the loop one person at a time. When Chatbase's founder posted a 20-second demo to 16 followers, it spread partly because he tagged the infrastructure tools he had used, and those companies reshared it to their large audiences (Decoded Startups). Borrowed reach beats a follower count of zero.

Your best growth channel at this stage is your existing users. Talk to every one of your first customers and ask two questions: how did you find us, and what nearly stopped you from buying. The first answer tells you which channel to double down on. The second is the exact objection to fix in your copy. Then ask, plainly, if they know one other person with the same problem. A referral ask in the first 30 days often outproduces cold outreach, because the person already trusts your customer.

Your first two weeks

A concrete plan you can start on a Saturday. The goal is not 100 users in two weeks. It is to find the one channel that produces a real user, so you can run it as a weekly loop after that.

• Day 1: Write your warm-30 list. Thirty people who genuinely have this problem. Message five of them individually today.
• Days 1 to 3: Pick three to five communities where your user already describes the pain. Install a session recorder like Microsoft Clarity on your site. Read 20 threads without posting.
• Days 2 to 7: Answer real questions in those communities with no link. Finish the warm-30 messages. Offer a personal setup call to anyone who says yes.
• Days 5 to 10: Watch your first users use the product on a call. Fix the top thing that trips them up. Rewrite your landing page around the words they used.
• Days 8 to 12: Send 10 to 15 personalized outreach messages a day to people who have posted about the problem. Reference their specific post. Never blast.
• Days 10 to 14: Post one honest build-in-public update. Ask every user so far how they found you and who else has the problem. Line up a launch only once billing and onboarding hold together.

Bottom line

Distribution is harder than building, and AI made that gap wider by making building trivial. But the first 100 users have never required a budget or an audience. They require you to go where the problem is already being discussed, recruit people by hand, treat your launch as a test rather than a finish line, and turn the users you win into the ones who bring the next batch. Pick one channel, run it every week, and judge it on paying customers. Do that for a month and 100 stops being a wall. It becomes a checkpoint.