I’m learning to code (with AI)

Since I started TCLA six years ago, I have long been constrained by my inability to code. This doesn’t just mean I can’t solve problems myself. More importantly, it means I have a limited perspective on what is possible to build. 

This is all about to change. 

Right now, my Twitter feed is filled with software engineers questioning the future of their work thanks to OpenAI’s upcoming o3. 

But, personally, there couldn’t be a better time for me to learn how to code because of how AI can help and answer questions along the way.

I already find my default is to ask questions to ChatGPT rather than Google.

What I want to develop is the foundational knowledge so I can build things, even if in practise I continue to hire developers for bigger projects. Long-term, it’s a bet that having this knowledge now will compound and that it’ll be a superpower to be like: ‘Oh I want this thing and I can build this in the way I want to’.

This is the big downside to not building things myself. I find I can’t make things the way I want them which is limiting when our edge is having a deep understanding of the problem we’re trying to solve.

And right now, it’s quite clear that we’re really early. Soon, pretty much everything about how we work is about to change. So there is an arbitrage to learning this stuff before it becomes mainstream.

So my goal is to build an MVP of our new commercial awareness platform for release on the 1st February 2025. I suspect having a concrete project will be the best way for me to learn. I hope to share my journey here along the way. 

For context on what I’m using right now: I’m currently using Cursor AI with Claude 3.5 Sonnet. I’ll build the first version of the app using Python. I am then using ChatGPT to ask questions when I get stuck.

I’ll share my journey here along the way. If you can code, please let me know if you have any advice and/or I’m doing some wrong(!)

Baby steps…