Why we built this
Research justice
is not a slogan.
It’s an engineering
problem.
Diseases that disproportionately harm Black patients, Indigenous patients, and communities of color are systematically underfunded and understudied. This is not an accident. It is a structural problem — written into funding priorities, trial enrollment criteria, and the careers of researchers who have never been able to get this work funded.
The human cost is not abstract. It is 80% of Black women developing uterine fibroids by age 50 while only 3.7% of published fibroid research includes race-stratified outcomes. It is women presenting to emergency departments with symptoms that were never studied in people like them, treated with protocols that were never designed for them.
Project UG is research justice infrastructure. We build AI agent systems that work neglected problems continuously — synthesizing evidence, identifying gaps, proposing hypotheses, and surfacing findings that belong in grant applications and clinical trials.
The evidence
The gap is not anecdotal. It is measurable.
higher lifetime fibroid incidence in Black women vs. white women
Baird et al. 2003 (PMID: 14583444); Wise et al. 2016 (PMID: 26739045)
earlier median fibroid onset in Black women — 28.3 vs. 35.1 years
NHANES 2015–2020 matched cohort, n=4,847, p<0.001
of NIH fibroid R01 grants include race-stratified endpoints (2015–2024)
NIH Reporter analysis — Project UG Cartographer, 2024
Black women in fibroid clinical trials vs. 40%+ of disease burden
Stewart et al. 2017 (PMID: 28903101)
The methodology
Seven agents. One research brief.
Built to close the gap.
Each research question is processed by a coordinated swarm of seven specialized AI agents. They do not simply summarize papers. They build a structured case — mapping the evidence landscape, identifying cross-disease analogues, running epidemiological pattern analysis, proposing mechanistic pathways, stress-testing every claim, auditing for ethical requirements, and synthesizing everything into an exportable research brief.
The Skeptic agent downgrades claims that are not adequately powered. The Ethicist agent flags historical consent context and community benefit requirements before any study proceeds. These are structural requirements built into every run — not optional.
The output is a confidence-scored research brief designed to be pasted directly into a grant application, sent to a patient advocacy group, or submitted to an IRB. Research justice infrastructure means the outputs have to be usable by the people doing the work.
Ethicist Agent — Why it exists
J. Marion Sims conducted foundational gynecological research on enslaved Black women without consent. That history is not a footnote.
Every proposed study design that emerges from Project UG must pass through the Ethicist agent before it reaches a research brief. The agent enforces three mandatory requirements: historical consent context in IRB submissions, community benefit criteria defined before data collection begins, and operationally defined race classification that acknowledges race as a social — not biological — construct.
This is not compliance theater. Research that extracts data from Black communities without delivering benefit back to those communities replicates the same structural injustice that created the research gap in the first place. The Ethicist agent is infrastructure — not an afterthought.
What’s next
Uterine fibroids are the beginning.
Every disease on this roadmap has a research gap score above 7.5 — meaning the burden on underserved communities is high and the research investment is critically misaligned. Project UG will deploy the same agent infrastructure to each one.
Uterine Fibroids
Gap score: 9.4/10
Keloid Formation
Gap score: 9.1/10
Hypertensive Kidney Disease
Gap score: 7.8/10
Sickle Cell Disease
Gap score: 7.9/10