The Recommendation Engine for Commodity Traders — Anekha Sokhal, Moshi
Moshi founder Anekha Sokhal, a former energy derivatives trader, explains building a real-time, recommendation-driven insight engine for traders, and what AI engineers and buyers told her after her talk.
Watch on YouTube ↗Charlie's summary
- ▶ 5:37
Moshi is the program Anekha wished she had as a trader: real-time insight across more data than a human can read
Anekha says Moshi processes data at a pace where it “has learned 300 new things” in the 4 seconds it takes her to say a sentence. It combines market time series, fundamentals, news, X and podcast audio, plus each user's private analysts, subscriptions and internal data. It cleans, tags and builds an ontology, then pushes updates and answers questions in about 30 seconds that would take days.
- ▶ 9:17
Moshi is not a new Bloomberg: it's Bloomberg, Claude and Netflix having a baby
Anekha explains the Netflix part as an internal recommendation system. A trader looking at one thing needs the second- and third-order effects right away, such as an energy equities company's move that indirectly hits oil. A news stream won't surface that. A connected graph that mimics a human making the link can, and she wishes she'd had one when trading.
- ▶ 11:48
If everyone can build their own dashboard, the hard problem becomes UI and recommendation, and Anekha thinks solving one solves the other
Anekha notes that anyone can now build a Claude dashboard, but getting someone else to adopt it is a different challenge. Platforms like Spotify and Instagram win by pushing data to you, which hooks you. Moshi's thesis is that if you solve UI you naturally solve recommendation, and vice versa. Swyx adds that recsys is hugely important yet barely attended at his conference.
- ▶ 15:21
Disagreement between sources is itself a trading signal, so show confidence scores rather than a single answer
Anekha says traders trust particular voices on X, such as commodity suppliers and LNG facilities posting status updates, but Moshi filters and attaches a confidence score. It also shows what else is happening around a claim. If Reuters, Bloomberg and the Wall Street Journal disagree, the disagreement is a data point: “is there an opportunity there because no one knows.”
Give your email to unlock the rest of the key points — for every episode.
Do more with Charlie
Free — sign in with your email, no password.