AI & Building
From Trading to Building with AI
What trading businesses and operational challenges taught me about building useful products with AI.
By Luca Cerullo · Published
Start with the work, not the tool
My work across trading businesses, education and international operations brought me close to the practical side of running a company: deciding what matters, coordinating people and carrying decisions through. That experience shapes my interest in AI. I want to start with a problem someone actually has, rather than a feature that happens to be possible.
For a founder, that means describing the work before choosing the software. Who needs help? Where does the process become difficult? What would a useful first version let that person do? These questions give a product a purpose and make it easier to decide which ideas belong in the first release.
Give an idea a roadmap
Codexa Code brings this direction into software: AI vibe coding that follows a roadmap autonomously from an initial idea to a complete, functional MVP. The value I am focused on is continuity between the idea and the work needed to deliver it. A roadmap gives that work a direction instead of leaving the owner with disconnected experiments.
The owner still has decisions to make. Define the intended user, the problem and the boundaries of the first version. Then review the working result and refine the direction. A functional MVP is a starting point for learning about the product; it is not evidence, by itself, that people will need or buy it.
Keep responsibility with the business
I exited Prop Number One in January 2026; the company later closed in July 2026. I exited Blue Chip Broker in January 2026, after its April 2025 South African acquisition; it was never launched as a standalone public retail brand. Those experiences remain part of how I think about execution. An ambitious plan does not remove the need to understand operating costs, ownership and the decisions that keep a business moving.
When I approach an AI product now, I want a clear next outcome and a way to review it. That might mean checking whether a workflow makes sense to its intended user or whether a first version addresses the original problem. AI can help carry the work forward; the business still needs people who question the result and take responsibility for what happens next.
Turn the next useful idea into working progress.
Bring a real problem, a clear audience and the next decision you want to make. Explore a hands-on product, operations or AI collaboration.
Let’s Build →