/
© 2026 RiffOn. All rights reserved.

Get your free personalized podcast brief

We scan new podcasts and send you the top 5 insights daily.

  1. Latent Space: The AI Engineer Podcast
  2. The Recommendation Engine for Commodity Traders — Anekha Sokhal, Moshi
The Recommendation Engine for Commodity Traders — Anekha Sokhal, Moshi

The Recommendation Engine for Commodity Traders — Anekha Sokhal, Moshi

Latent Space: The AI Engineer Podcast · Oct 22, 2026

Moshi combines LLMs and recommendation systems to surface real-time, second-order market insights for energy and commodity traders.

Graph recommendation engines beat news feeds for trading risks

Anekha Sokhal explains that commodity traders miss indirect risks when relying solely on traditional news feeds. Moshi uses an internal graph recommendation engine to connect secondary and tertiary market impacts, like an energy stock move affecting oil, pushing insights traders would otherwise overlook.

The Recommendation Engine for Commodity Traders — Anekha Sokhal, Moshi thumbnail

The Recommendation Engine for Commodity Traders — Anekha Sokhal, Moshi

Latent Space: The AI Engineer Podcast·a month ago

Management overburdens solo AI engineers to replace ten-person teams

Anekha Sokhal observed that AI engineers at the conference felt overwhelmed by executive demands to single-handedly construct massive systems. Leaders increasingly tell individual developers they are the whole team now, expecting one person to deliver complex architectures that previously required ten specialized software engineers.

The Recommendation Engine for Commodity Traders — Anekha Sokhal, Moshi thumbnail

The Recommendation Engine for Commodity Traders — Anekha Sokhal, Moshi

Latent Space: The AI Engineer Podcast·a month ago

Proprietary financial data forces startups to build in-house models

Anekha Sokhal notes that Moshi develops proprietary vision language models rather than routing user documents through third-party APIs. Institutional trading clients refuse to upload sensitive, unstructured PDF data unless the startup guarantees internal data processing and prevents outside vendors from touching confidential records.

The Recommendation Engine for Commodity Traders — Anekha Sokhal, Moshi thumbnail

The Recommendation Engine for Commodity Traders — Anekha Sokhal, Moshi

Latent Space: The AI Engineer Podcast·a month ago

Disagreements between news sources provide valuable market data

Anekha Sokhal highlights that when major news outlets like Reuters and Bloomberg report conflicting facts, like ship counts, the discrepancy itself represents actionable signal. Conflicting reports reveal underlying market uncertainty, giving commodity traders unique arbitrage opportunities rather than just being discarded as noisy data.

The Recommendation Engine for Commodity Traders — Anekha Sokhal, Moshi thumbnail

The Recommendation Engine for Commodity Traders — Anekha Sokhal, Moshi

Latent Space: The AI Engineer Podcast·a month ago

Push interfaces beat dashboards for enterprise software adoption

Anekha Sokhal points out that building customized dashboards usually fails because new users will not adopt another person's mental model. Instead of relying on manual pull requests, software should mimic consumer platforms like Spotify by pushing hooked data directly into daily user workflows.

The Recommendation Engine for Commodity Traders — Anekha Sokhal, Moshi thumbnail

The Recommendation Engine for Commodity Traders — Anekha Sokhal, Moshi

Latent Space: The AI Engineer Podcast·a month ago