Pharma·CT8K
A live intelligence product, designed and built end to end by one person.
The brief
Two public datasets, no shared keys, both messy. Clinical-trial records (the AACT database, 579,000+ studies) and SEC 8-K filings (12,500+ filings across 766 companies). Separately, each is noise. Cross-referenced and scored, they surface material developments before they are formally disclosed.
The product turns that overlap into a single monitoring layer: a signal pipeline, confidence scoring, a therapeutic-area Crowding Score, an interactive density atlas, and a structured feed. It updates on weekday evenings.
Role & scope
Solo and full-stack, end to end. The data comes from public sources (ClinicalTrials.gov via AACT, SEC EDGAR); everything else is Andrew's: the product, the UX, the intelligence model, and the build.
It runs under Craton Meridian, his AI Integrity firm, as its pharma intelligence surface. What is live and public here is the public-data layer only, shared as a case study with permission. Craton Meridian overlays proprietary data on the same system for paying clients.
Process · the decisions
01 · The signal pipeline
The live product itself documents the technical methodology in detail, so the case study keeps this short. What is worth saying plainly: building it required serious subject-matter research and comprehension across both domains, the clinical-trial data and the SEC filings, with no shared keys between them. Andrew's own AI-assisted deep-research and validation process is what made that depth of understanding possible.
02 · Scoring confidence honestly
Both public sources are noisy relative to what the product needed from them, so the confidence-scoring model went through a lot of tuning. Andrew describes it as a slow process, "like a sculptor gently shaving off the next slice," careful at every step not to cause destructive data loss.
03 · The Crowding Score
The Crowding Score is where the whole project started, not the other way around. Craton Meridian wanted to understand the pharmaceutical landscape better, who the real movers and shakers were. That question is the genesis of Pharma·CT8K as a whole. Ranking therapeutic areas by growth, density, sponsor diversity, and mechanism diversity is how the atlas answers it at a glance.
Outcome
It's a completely autonomous, self-updating dashboard, live and free for public consumption.
Media
