Forem

huangyongshan46-a11y
huangyongshan46-a11y

Posted on

I Built a SaaS in a Weekend Using a $49 Starter Kit — Here's Exactly What Happened

#saas #nextjs #indiehackers #webdev

I Built a SaaS in a Weekend Using a $49 Starter Kit — Here's Exactly What Happened

Friday evening. I had an idea I had been sitting on for two weeks. A simple AI-powered tool for a specific niche — nothing revolutionary, but something people in that niche would genuinely pay for.

I gave myself the weekend. Here is what actually happened.

Friday Night: Setup (2 hours)

I picked up LaunchKit — a Next.js 16 SaaS starter kit I had been eyeing. $49. I have spent more than that on a dinner out.

The purchase-to-running-locally pipeline was faster than I expected:

  1. Download the zip from Gumroad
  2. npm install
  3. Copy .env.example to .env.local
  4. Set up a Postgres database on Railway (took 4 minutes)
  5. Run npx prisma migrate dev
  6. npm run dev

Twenty minutes later I had a running Next.js app with auth, a working database, and a Stripe checkout flow stubbed out.

The first thing I did: actually read the code. Not the docs — the code. LaunchKit is clean enough that the code is self-documenting. App Router, proper file structure, everything where you expect it.

By midnight Friday I had:

  • Auth working (Google + email magic links via Auth.js v5)
  • Database running with my custom schema added
  • The basic shell of my product

I went to sleep feeling good.

Saturday: The Build (10 hours)

This is where it got real.

Morning: Core feature

The AI integration was why I chose LaunchKit specifically. It comes with the Vercel AI SDK already wired up — not just installed, but with example API routes and streaming response patterns.

My product needed to call OpenAI with a specific prompt, stream the response to the client, and save the result to the database. With LaunchKit's AI boilerplate already in place, I had this working in about 90 minutes.

// This pattern was already set up — I just customized the prompt and response handling
export async function POST(req: Request) {
  const { input } = await req.json();
  const result = await streamText({
    model: openai('gpt-4o'),
    prompt: buildMyCustomPrompt(input),
  });
  return result.toDataStreamResponse();
}
Enter fullscreen mode Exit fullscreen mode

Having this scaffold meant I could focus on my business logic instead of figuring out streaming.

Afternoon: Stripe and pricing

This is usually where I lose hours. Not this time.

LaunchKit has Stripe checkout and webhook handling already built. I needed to:

  • Update the product/price IDs in .env
  • Customize the pricing page (already a component, just swap the copy)
  • Test the checkout flow

Two hours. Including testing with Stripe's test mode cards.

The webhook handler for subscription events was already there — I just extended it to update my database when a user subscribed or cancelled.

Evening: UI polish and the hard parts

Honest admission: the parts that took longer than expected were the parts unique to my product. The core logic, the edge cases, the UX decisions.

LaunchKit gave me everything surrounding the product — auth, payments, database, AI plumbing. What it cannot give you is your actual product logic. That is still on you.

I spent Saturday evening deep in my core feature. Fewer lines of code than I expected. More thinking than I expected.

Sunday: Polish, Deploy, and Launch (8 hours)

Morning: Deployment

Vercel deployment took 15 minutes. LaunchKit's structure is Vercel-native — no config needed beyond setting environment variables.

One thing I appreciated: the environment variable setup is well-documented in the kit. It is clear what each variable does and where to get it. No hunting through the code to figure out what NEXTAUTH_URL should be.

Afternoon: Domain, email, the small stuff

I pointed a domain at Vercel, set up email via Resend (LaunchKit uses Resend for transactional email — already configured, just needed my API key), and went through the whole flow as a user:

  1. Land on marketing page
  2. Sign up
  3. Use the product
  4. Upgrade to paid
  5. Cancellation flow

Found three bugs. Fixed them. One was mine, two were edge cases in the auth flow I had not considered.

Evening: Soft launch

I posted in two subreddits and sent an email to a small list. Not a big launch — just testing whether anyone cared.

12 signups by Sunday night. 2 paid. $49 revenue on day one — basically covered the cost of the kit.

One Week Later

Some honest numbers:

  • Revenue: $147 (3 paid customers)
  • Signups: 34
  • Conversion rate: ~9%
  • Hours spent building: roughly 20

Not life-changing. But it is a real product that real people are paying for, and I built it in a weekend.

What LaunchKit Gave Me

Time savings:

  • Auth setup: probably 3-4 hours saved
  • Stripe integration: 4-6 hours saved
  • AI SDK setup: 2-3 hours saved
  • Database + ORM config: 1-2 hours saved
  • Email setup: 1-2 hours saved

Conservatively, 12-16 hours saved. At any reasonable hourly rate, that is well above the $49 cost.

Stack quality:
Everything is on the latest versions. App Router, Auth.js v5, Tailwind v4, Prisma. I did not have to make any uncomfortable compromises with the stack.

Code I can understand:
This matters more than it sounds. Some starter kits are so abstracted that you cannot figure out where things live. LaunchKit is readable. When I needed to customize something, I could find it.

What It Does Not Give You

Your product. Obviously. But also:

  • No i18n — if you need multiple languages, you are adding that
  • No multi-tenancy — single-user model out of the box, team features need work
  • No admin panel — you will need to build or add that
  • No analytics — add your own (Plausible, PostHog, whatever)

For my use case, none of these mattered. For some products they would.

Would I Use It Again?

Already am. I have another project in progress using the same kit.

At $49, LaunchKit is not a gamble. If it saves you even four hours of setup time, it has paid for itself multiple times over. And if you use it to ship a product that makes $100, you are already 2x on the investment.

The stack is genuinely modern. The code is clean. The AI integration is real, not bolted on.

If you are thinking about it — just buy it. The friction of setup is the most underrated killer of side projects. Removing that friction is worth $49.

LaunchKit on Gumroad: https://yongshan5.gumroad.com/l/xckqag

Code preview (browse before buying): https://github.com/huangyongshan46-a11y/launchkit-saas-preview

What are you building? Drop it in the comments — always curious what people are working on.

Top comments (0)