A story about AI tools, and how they will change the world.

Sep 30 - 2024

A good friend recommended me to try Cursor.sh, so I did.

I was not ready for what happened next.

It all began with a simple idea I'd been pondering for a while. Whenever I needed to check if a text was correct, I had to go through a cumbersome process: find my ChatGPT tab, log in if disconnected, start a new chat, type "fix typo:", paste my text to fix, and then copy the corrected text back. It was far from efficient.

I knew I could create a simple website using OpenAI APIs that would accomplish the same task without needing to write "fix typo:" every time—a straightforward hack. Of course, I never found time to start this pet project. That is, until I decided to give Cursor.sh a try.

Cursor.sh is another IDE, but it's a fork of Visual Studio, which is (almost) everyone's favorite. Visual Studio best extension is GitHub Copilot, a tool that autocompletes the code you're writing.

Copilot is so impressively good that it feels like it's reading your mind. It accurately predicts the next function you're going to write. With just a press of the tab key, it writes it for you.

There are also ChatGPT and Claude, both excellent at writing code. However, you have to copy and paste their output into your editor.

But Cursor.sh takes it to another level. It optimizes the integration of Large Language Models (LLMs) even further. The LLMs have access to more context, like your full codebase or error messages if your code crashes. You can ask Cursor.sh to do complex tasks like "Move the business logic of this function to the backend," and it will suggest modifications across multiple files for you to review before accepting them.

The point isn't that Cursor.sh's AI is perfect—it's not. Mistakes happen often. The key is that the AI knows enough to get you started. You receive your first reward faster, and it assists you at every step. Rewards continue to come as you progress. This reduces frustration because you're never stuck alone. You can always ask the AI for help or explanations about what's happening, what's expected, and how the code works. Every AI failure becomes a learning opportunity. It elevates "learning by doing" to a new level. With will, the rest follows. It's never been easier.

Here is the result of this experiment: fix-typos.com

The AI generated 95% of the code.

Polishing the UI consumed 80% of the development time.

But most importantly, I've never felt more empowered.

So here are multiple predictions based on what I learn last week.

• In 5 years, if you ask someone on the street about their AI "Aha" moment, it won't be ChatGPT. It will be tools with a narrower scope that empower them in a much deeper way. ChatGPT giving better answers than Google is nice, but when you suddenly feel you can do 10 times more of what you love, it's something else entirely.
• The tech industry will be the sector most disrupted by AI, by far.
• Software engineering schools will have to reinvent themselves. Learning by doing will become the norm.
• A new generation of bootstrapped startups will rise, with co-founders no longer hiring full-time employees.
• The efficiency gap between good and bad software engineers will widen dramatically, not narrow.
• UX will become the main differentiator for many products.