Launching today

Indie Founder AI Launch System
Idea to launch in 8 gated stages, with kill criteria
1 follower
Idea to launch in 8 gated stages, with kill criteria
1 follower
A step-by-step system for Claude, ChatGPT & Cursor: 8 gated stages from idea to launch, each with a template, checklist, tool-specific prompts and a GO/CHANGE/KILL gate. Worked example included. PDF + Notion-ready Markdown.






Hey Product Hunt, I'm Samuel from Midaz.
You might know Midaz from client web and AI infrastructure work. Lately I've been turning what I've learned into products for founders.
I've been building businesses and technology for more than 15 years, but AI changed the way I think about building products.
Over the past few years, I've spent an increasing amount of time experimenting with AI agents, automation, knowledge systems, and AI-native products. What started as curiosity became a much bigger realization: AI is not just another software feature. It is becoming a new way to build, operate, and scale businesses.
I didn't want to create another AI wrapper or chatbot. I wanted to solve a problem I was experiencing myself. So I started building it, testing it, breaking it, rebuilding it, and putting it into real-world use.
The problem: a lot of indie founders already have Claude or ChatGPT open in one tab (often Cursor too) and plenty of ideas. What's missing is the order of operations. When do you stop researching? What do you charge? What do you not build? When do you kill the idea?
So I wrote that order down as a system with gates, not a list of prompts:
Setup + 8 stages: idea & problem validation โ customer research โ offer & positioning โ pricing โ MVP scope & build โ landing page โ pre-launch & launch โ post-launch review.
Every stage has the same structure: goal, inputs/outputs, steps, a fill-in template, prompts, a checklist, and a GO / CHANGE / KILL gate. Each stage writes one file that later stages read, so your landing copy comes from your interview notes and your MVP scope comes from your offer stack. You write kill criteria before the evidence comes in, which makes it harder to quietly move the goalposts.
What's inside:
- 45 prompts in 77 tool-specific versions. Claude, ChatGPT, and Cursor each get their own, and every prompt tells the AI not to invent stats, quotes, or competitors
- Claude Project and ChatGPT Project instructions, plus a Cursor rule file
- A fictional micro-SaaS (Quellbird, payment reminders for freelancers) worked through every stage. It's clearly labelled as fictional
- 80+ page PDF + Markdown you can import into Notion or keep in your repo
You only need Claude or ChatGPT. Cursor is for the build stages if you code.
If you're not sure yet, I made the validation part free as the Micro-SaaS Validation Mini-Pack: https://getmidaz.gumroad.com/l/v...
Launch week: $29 with the launch code (applied automatically through this page's link) through Wed Oct 14, 11:59pm PT, then $39. 30-day no-questions refund.
I believe the best products are shaped by people who actually use them. So: which stage do you usually get stuck on? If you try it, tell me what feels useful, what feels unnecessary, and what you would build differently. I'll be here replying all day.