Over the next five days I am going to share with you my most valuable findings I've discovered over the last 6+ months building my own SaaS.

I've learned a lot. The development space is changing faster than ever before. Turns out, building a product is not the only challenging part.

PS, it's Caleb from CodeBreakthrough here. I'm sending my emails from a new email provider, so if this landed in spam please do me a favor and move it to your inbox / mark it as not spam to make sure you get my new stuff!

The Challenge of Building Today

With the help of AI it's now possible to create software faster than ever. This introduces two main challenge...

First, AI usually only gets you 80% of the way to a final product quickly. When you're a beginner with agentic development you will get you a great prototype, but getting the rest of the 20% takes the most time and skill. In my case, I spent a month or two building out my main features and four months improving them. If you don't have the skill or patience to get those final improvements you're going to end up with mediocre software.

The other major thing is that your software will blend in with a flood of other mediocre software with so many new apps being created. Anyone can replicate features and create apps, so it becomes much harder to build software that stands out.

If you want a successful SaaS you have to create something people want and are willing to pay for. Another thing you should be aware of is that AI is better at replicating behavior (anchoring to existing examples) than generating something new, meaning people can copy your features quickly. How can you create anything unique if it can be copied with AI?

How can you fix this? You can work to identify a moat (competitive advantage) that is beyond just simple features. Perhaps it's the quality of features or the speed at which you can solve custom problems. Or maybe it's the community and resources you build AROUND the software, such as next-level support.

How to be better than everyone else

If you learn to properly build with AI you have higher leverage than other developers. Proper use of agents allows you to build better features, faster.

The first thing you can do is learn to increase your leverage, getting more from the same input (translation: become a more skilled operator of agentic AI). This goes beyond prompt engineering in to highly predictable autonomous AI workflows.

When used properly, agents can help you achieve results that are better and much faster. But if you don't know what you're doing it's the opposite effect. It just amplifies the output if that makes sense.

For example, you could spend a day learning agentic systems, and as a result be able to have Claude write for hours instead of asking for input every few minutes.

From here you work on your personal skills of focusing on higher-challenge problem. For example: How can we design our codebase better? What performance improvement can we make? What UX flows are killing customer success? A lot of engineers have a hard time stepping out of "feature mode" where they see progress how many issues they're able to fix, while completely failing to see the bigger picture.

Finally, learn product, marketing, and sales. What makes a good product? What do users want and need?

Imagine for a moment the absolute extreme: that anyone can INSTANTLY generate software that does exactly what your software does. What would you do? You have to think about the messaging and positioning, how to market, and how to solve real problems. Are you able to get someone to give you their credit card information? If not, get back to the drawing board.

What will we look at over the next 4 days?

This email was an introduction, to help you realize how agentic is disrupting software development.

First, we need to know our tools. So we'll cover the essentials you should know.

Done poorly AI will just get you a sloppy prototype, so once we have our tools figured out thing we will explore how to use these tools to get quality output from AI.

Then, we'll learn about permissions, security, and autonomous AI. Our goal is to give AI as much freedom without putting ourselves at risk.

Finally, we'll look at larger agentic workflows where I will show a systematic approach to utilizing subagents for building larger features or improving your existing software.

By the end we will have a much stronger understanding of agentic AI.

After polling my audience and also hearing some questions, I realized that maybe we need some more foundational intro content before jumping right in to advanced agentic ai patterns. So, this course is designed to build the gap from beginner who knows a bit about AI to someone thriving in agentic engineering.

Please reply if you enjoyed this or if it was helpful. Any questions? Comments?