CodeBreakthrough - Caleb Curry

Today I'm wrapping up some content I've been working on and announcing something new.

As a background, I created some content around AI for software engineers. I couldn't finish the final section because I had a blocker...My goal? Utilize AI to launch a product and business that was high quality and profitable. The problem was scope creep...what started as an idea to launch a quick simple app on the app store turned in to a full blown platform that has taken months of effort to build.

Well, this software is now early access and is already supporting paying customers, so I finally finished recording that content to push it live this week.

So we're done, right?

Not exactly. I got a product launched but I don't want to stop there because I just scratched the surface of what's possible.

My approach to development at the start of the app was completely different compared to my approach now. This is because throughout the entire process I was learning new things, trying new tools, and approaches. Plus, the AI landscape is constantly evolving.

Here is a visual of my AI progression. There was legitimately a point in the process where I said "I'm done with AI" and wrote mostly everything by hand. But, I kept learning, and eventually something clicked.

​

The biggest unlock? ...agentic development (done properly)

The idea behind agentic AI is that we have an AI working towards a goal. nothing too crazy, but agentic doesn't automatically mean autonmous...we can have agentic AI that still requires constant monitoring and guiding, slowing us down, or we can set things up to actually increase the quality of our work and help us do it faster.

My new way of working is allowing me to build and support my entire business, development pipeline, and products being used by real users.

But most people are using AI wrong. They prompt over and over again, failing to get agents to run for longer periods of time. Or end up with spaghetti code that is impossible to test or edit yourself.

  • What if I told you it was possible to use AI to build better software without creating vibe coded AI slop (features that suck so bad you'd rather just code yourself)?
  • What if the AI actually enhanced our capabilities and allowed us to think about higher level challenges (like building a product that people will actually use)?
  • What if we could get an AI to build for hours, without interruption, creating a result that actually works as you planned?

Trust me when I say I was hesitant to jump on the AI hype train...that's because so much out there is marketing garbage with very little clarity on how to actually use AI properly. It's something that has taken me months of learning and experimenting, and I want to share it all with you.

So, let me wrap up by asking a few simple question...

Who are you?

  • A dev wanting to upskill for my dev career
  • a builder wanting to build products end-to-end
  • A business / founder looking to utilize AI

How are you using AI?

  • Almost none - Basic prompting + copy/paste
  • Some - I use claude/codex/copilot like a pair programmer
  • power user - Fully agentic... custom skills, subagents, hooks, etc

What do you want to accomplish?

  • Learn coding faster, better (AI for enhanced learning)
  • Excel at my job / career (AI for job security)
  • Simplify and automate stuff (AI as an automation tool)
  • Utilize AI to grow my business, launch products, or make money

So if it's not clear, Actually reply. You can copy and paste the questions with answers or write something completely custom. This isn't just a filler for me to wrap up the email... the replies go to my inbox and I read them and reply to most. So yea, hit reply and answer the questions and I'll share my thoughts. (actually me, not an AI, lol)

Caleb

​Unsubscribe | Update your profile | Caleb Curry 500 Westover Dr #17334, Sanford, NC 27330-8941