About Underground Commerce.

About Underground Commerce

Underground Commerce is a research project run by Lexora Labs, exploring whether structured synthesis of media signals can materially improve human decisions in domains of high informational uncertainty. The product is a reading aid, not a forecaster — when narrative and money disagree on a topic, the divergence layer surfaces the gap and offers a defensible theory of why, so a reader can form a sharper view faster than they could from unaided media consumption.

The intellectual lineage is older than the model behind it. Kira Radinsky and Eric Horvitz published Mining the Web to Predict Future Events in 2013, describing a system that learned from 22 years of New York Times archives to forecast disease outbreaks, deaths, and riots. The retrieval, abstraction, and grounding mechanics — now implemented with LLMs rather than classical NLP — are the technical template Underground Commerce is built on. The research question is adapted: not whether text mining can predict events, but whether the same machinery can produce a useful decision-support read.

Two flagship domains. Business strategy is live today across eight tech-business categories — capital markets, labor signal, regulatory pressure, AI deployment risk, and four more pending source coverage. Career transitions is the planned second domain — a DOORS-inspired tool for job seekers, drawing on O*NET occupations, BLS wage data, and CareerOneStop demand signals to surface adjacent careers worth considering. Both domains share the same shape: two independent reads, structured disagreement, explained gap.

Why we're building it in public

Most products like this ship a polished surface and treat the methodology as proprietary. We're going the other way. The methodology is the product, and being able to read how it works is part of what makes it useful. When the v2 pipeline ships with backtested results, it'll come with the data and the limitations. When something's wrong, we'll say so.

What we hope it becomes

The dashboard is the front door. The reports — when they ship — are the interpretation: analyst-written briefings that connect dots across multiple categories, structured as summary, sections, and a watch-list of leading indicators.

We're not trying to be Bloomberg, and we're not trying to be a crystal ball. We're trying to be the thing you check once a day so you don't have to read everything else — and the thing you turn to when you're staring down a decision (a strategic bet, a career move) and need a structured read of the noise.

Contact

For partnerships, press, or questions: [email protected].

For product feedback: same address. We read everything.