I think you’re asking the wrong question. Startups (and companies in general) exist to solve a need. AI is a tool that solves a set of problems. Ymif you can solve a wide population’s problems without AI, then do it.
An algorithmic trading hedge fund. We can outperform the benchmark while keeping fees bare minimum because all our admin is outsourced to LLMs and agents. What a world we live in.
One of the most successful quant traders in history, Jim Simons studied topology in the 60s. It’s rumored he used deep neural networks in his trading before they were cool. This post really brought the two together for me in a way I didn’t understand before.
I typically work within Jupyter Notebooks since it’s easier to build tutorials with text and markdown. But I agree: even looking at the code output I can tell almost immediately it’s made up.
That is some rare foresight. Typically, this is not how it goes. Usually, the pressure to keep adding new features is too high, the codebase too large, and the desired functionality too amorphous for any sort of refactoring or rewrite.
It was a very sudden mind shift for me. All of the sudden one day I said f* it. If I want this thing to exist, I need to focus on shipping. Not building.
The next day I hired up the freelance team.
Some people just like to build. Others want to run a product.
