Week one of my 30 projects in 30 weeks challenge didn’t go as planned. I spent most of it sick, but somehow managed to tackle something I’d been putting off for years: building my own boilerplate.

Thankfully, I use a lot of speech to text and managed to get a lot done during the times when I was actually awake. I worked on an all-in-one Next.js boilerplate. I had built many boilerplates off and on for the past maybe four or five years across multiple different frameworks, but I never finished any of them. Things kind of always got in the way and, you know, cue out the excuses.

I thought ahead of this 30 and 30 week challenge and I also thought about how expensive AI is getting and how my income is down to zero. I realized that probably the smartest thing to do right now would be to create a bunch of templates for web applications and mobile applications so that if in the next six months to one year the cost of AI spirals and I’m still in a position where I could maybe afford some of it but not too much, I want to be able to still have something to start with where I can maybe use a cheap model and then code the rest by hand.

🤔 Why I Finally Built My Own Boilerplate

For years, I kept spinning up projects in a way that wasn’t efficient. I kept doing it from scratch and reinstalling the dependencies and then having to redo the structure and the database. I finally realized that a boilerplate or a starter kit would be better.

I had tried out many boilerplates and starter kits in the past, but it never really served my exact purpose. A lot of the things that they had were things that I could have easily done myself. What I really needed was something to help quickly start

• a blog
• personal website
• directory
• sales page
• a micro saas
• a waitlist with payments
• the ability to quickly swap different database or auth providers for cost savings

I just wrote a list of things that I enjoyed from other boilerplates and I wrote a list of things I wish boilerplates would make easier for me as someone that has so many ideas. That’s how I decided on the Namost Next Kit.

🧠 What I Learned About Development Workflow

The bigger lesson this week wasn’t about the boilerplate itself. It was realizing my development workflow was a mess.

I used to be the person who stuck to a monorepo and avoided branches altogether. I thought git worktrees and branching were overkill for a solo developer, something only big teams needed. That mindset worked fine until I started working with AI tools and agents. They’re amazing for speed, but they’re also chaotic. I’ve even had AI push changes directly to production—thankfully only on test projects.

That’s when I noticed how people like Kieran Klaassen run multiple Claude code agents at once with proper branching. Meanwhile, I was still on one branch, no worktrees, with agents editing the same files and creating endless conflicts.

It hit me that my “just roll it back” approach wasn’t going to cut it anymore. With multiple agents and tools in play, things can spiral quickly. They help you move fast, but you need guardrails to keep the chaos from breaking your projects—or worse, introducing security issues.

🚦 Every Developer’s Workflow: Development to Production

Here’s what I realized every developer should have in place, especially if you’re working with AI tools:

Three Separate Environments

• Local: your laptop with test data and test keys
• Staging: cloud copy that mirrors production but with safe test data
• Production: live site with real users and real keys

Branch Strategy That Actually Works

• Feature branches for daily work
• Dev branch that auto-deploys to staging
• Main branch that auto-deploys to production
• Never merge to main without testing in staging first

Environment Variables Done Right

• .env.local for your laptop
• Staging environment variables in your hosting platform
• Production environment variables in your hosting platform
• Never share databases or auth apps between environments

Git Worktrees for Efficiency

• Check out multiple branches simultaneously
• Work on a hotfix while keeping your feature branch intact
• No more losing context when switching between urgent tasks

The Safety Net

• Always test sign-in, basic CRUD, and key flows in staging
• Keep previous builds ready for instant rollback
• Set up basic error tracking and logs

This isn’t about being fancy – it’s about not breaking things when you’re moving fast with AI assistance.

---

p.s. I’m still quite sick, and wrote this with speech to text. Sorry for any mistakes.🙏🏾💜