Hello (Commodities) World!

Why this Site Exists

My fiancé lives and breathes AI, and being around someone like that rubs off on you. Now that I have a bit more free time, I’ve started exploring AI more deeply through the lens of my own industry: commodities.

By now, using AI is not new to any analyst. In my previous role at a hedge fund, I was already using AI heavily in my day-to-day work: to rapidly run ad hoc analysis, publish dashboards, code supply and demand balances, etc.

The Core Question: Can AI Agents take on real Commodities Analyst workflows?

So far, AI has sped up my work tremendously, but I (the human) was still the one doing the tasks. The question I’ve been curious to explore is whether, beyond boosting execution speed, there are parts of an analyst’s day that an AI agent can handle independently.

More specifically, can AI independently do something more useful than morning news summaries or web scraping? Of course, the direction, analysis, and judgment still belong to the analyst, and any AI output must be checked by a human. But within those boundaries, are there parts of an analyst’s day where an AI agent can take on a real commodities workflow: e.g. work through messy external information, update a model, and explain its reasoning in a way an analyst would trust?

This was something umimaginable a few years ago, but as models improve, the boundary of what AI can do keeps moving. Tasks that felt out of reach not long ago are starting to become possible. I think of it like rising water: as model capability increases, more and more tasks fall within reach.

AI will inevitably become part of commodity workflows. This blog is about figuring out where it fits best, how to use it well, and how to evaluate if it is doing a good job.

Thanks for stopping by!