@Charles Kim My rule inperfect results beats perfection. Just get stuff done.

Welcome to the skool. Is there anything I can help.you with?

Like they use to say in the old days. I'll add you to my Rolodex. What type of project you've built?

@Carlos Jimenez Great question. For me, the answer is: it depends on what I’m building and who I’m building it with. I use a mix of tools depending on the workflow: Rocket.new, Claude Code, Hermes, Codex, Replit, Base44, and Perplexity Computer. I don’t really think of them as “one tool to rule them all.” I think of them more like a production stack for different stages of building. For example, I’ve built custom agents and skills that let my non-technical, domain-expert business partner create apps without needing to understand the technical details. Then I set up the engineering side: GitHub workflows, unit tests, integration tests, functional tests, and QA checks so we’re not just generating demos — we’re trying to create something that can actually survive contact with users. A big part of my process comes from working at Amazon Web Services for 7 years. I bring in lessons from that environment around quality bars, production readiness, operational excellence, and customer obsession. So before I treat an AI-built app as “done,” I usually run it through things like: - Production readiness assessment - QA / quality bar review - Unit, integration, and functional testing - UX and UI audit - SEO audit - Security / reliability sanity checks - “Would a real customer trust this?” review The way I look at it: AI tools are great at helping you move fast, but the real advantage comes from pairing them with strong judgment, domain expertise, testing, and a high quality bar. As for as models, I have the $200/mo Claude Code, Codex and I prepaid a yer for perplexity computer. I use Codex via auth for OpenClaw and Hermes. I forgot to mention when I run my production readiness assessments I use codex to assess the other models. I use Claude Code when assessing codex code. I've really interested in OpenRouter fusion api to get Fable level of output. I'm in the process of going live with a sports app using the perplexity method. It's web base for now and I'm kicking off a private beta with 1 high school coach and two professional Canadian Football League customers.

Wow, you hit on a lot. Lets start with domain knowledge. Yes, I found you can gain domain knowledge from your past experience or select a co-founder with domain knowledge. You can partner with a potential customer that has the domain knowledge. Collaborate with them to build a MVP. You want leverage their domain knowledge to help you design a working MVP thaat solve a problem they have and by extension the niche. Even better find a leader in the niche that has an active following with at least 1k members. Then when you solve their problem interview them and place the interview on your socials and their socials. I'm sure you should be able to get at least 10 customers. I've done both. In my Complete Performance business, my co-founder is an certified professional, created CEUs for other professionals and mentor new certified professional,. We don't get a lot of customers yet, but he got invited to give a talk at an university and we were able to get a mentorship customer and a design partner for an app with are going live next month. My partner met a clinic owner and we were trying to sale him a website but after talking to him. He had an idea for an app for his 19k member instagram channel. We are doing a 50/50 deal on the app. His knowledge and mabual system and I will turn it into an app.