Kupkaike

Hey IH 👋

I’m a solo founder from Quebec, Canada. By day I’m an industrial project manager. By night I code. No CS degree, no funding, no co-founder.

I just launched Kupkaike — an AI platform that finds profitable digital product niches and creates the products for you.

Here’s why I built it:

I wanted to sell digital products but I kept hitting the same wall: how do you know which niche is actually worth building in? I’d spend days scrolling Gumroad, reading Reddit threads, guessing at demand. Then I’d create a mediocre PDF that nobody bought.

So I built a market intelligence engine (MIOS) that does the research for me. It scans Gumroad, Etsy, Shopify, and Amazon KDP with 10 analysis modules: demand, competition, trend velocity, monetization depth, risk, and more. Every scan pulls from real marketplace data — not AI guesses.

Then I added a product generator on top. Type a niche → get an 8-chapter, 10,000+ word premium product with cover image, sales copy, voiceover narration, and Pinterest pins. The whole pipeline takes about 10 minutes.

I launched 3 days ago with nothing but a Reddit post and a few tweets.

14 visitors. 2 paying customers. $50 in revenue.

I know it’s tiny. But a 14% conversion rate on cold traffic with zero brand recognition tells me the problem is real.

The business model is credit-based — users buy “cupcakes” (1 cupcake ≈ $1) and spend them on scans, product generation, and marketing assets. No subscription. Average margin is 80%+.

What I’ve learned so far:

1. The niche scanner is what hooks people. They come for the intelligence, they stay for the product creation.

2. Showing real data matters. People are sick of AI tools that hallucinate market data. Real marketplace scanning is the differentiator.

3. Credit-based pricing removes friction. People buy $10 to test, then come back for $40 when they see the output quality.

What I’m working on next:

∙ Public niche pages (SEO play — every scanned niche gets a shareable page)

∙ AI agents that find leads and generate marketing content automatically

∙ Distribution packs (export ready-to-publish files for Gumroad/Etsy/KDP)

Tech stack: Next.js, PostgreSQL, Prisma, Claude by Anthropic (Opus/Sonnet/Haiku), OpenAI (DALL-E + TTS), Stripe, DigitalOcean. Built almost entirely through Claude Code.

Would love feedback from the IH community:

∙ Is the credit-based model right or should I consider subscriptions?

∙ What would make you try a tool like this?

∙ Any red flags you see?

Happy to share numbers, architecture, costs — totally transparent.

kupkaike.com