Skip to main content
David Ramirez
← All projects

Active, recurring shipped deliverable

GenServe.AI: The Shortlist

A weekly healthcare AI newsletter curated from 100+ articles down to a human-reviewed top 10.

Organization

GenServe.AI / Future & Health Institute

Role

AI Scholar Intern, Summer AI Scholars Program, 2026 Cohort

Period

June 2026 – Present

Status

Active, recurring shipped deliverable

The Shortlist is a real, dated, weekly newsletter (for example, Issue 06, published July 2026), not a concept or a prototype. David works on the newsletter automation project as a member of the 2026 Scholars Cohort, alongside fellow scholars and mentors, not as its sole builder.

The problem

Finding good healthcare AI articles for a weekly newsletter used to mean manually checking news sites, journals, and social media, then judging each one by hand: slow, inconsistent, and easy to let weak stories through or miss strong ones.

The goal

Build and operate a system that collects healthcare AI news and research automatically, scores it consistently against an agreed quality rubric, and hands a human editor a short, high-quality list to make the final call from.

My role

As a Scholar in the 2026 cohort, I work on GenServe.AI's Top 10 newsletter automation project, the pipeline that collects, scores, and shortlists healthcare AI articles each week, alongside fellow scholars and mentors, and I contribute to the research, curation, and healthcare-audience framing behind each issue. It's a team-built system I work on and operate within, not one I built alone.

Process

  1. 1The system automatically pulls new articles weekly from about 10 trusted healthcare IT, AI-industry, and academic sources, including PubMed and arXiv.
  2. 2Each article is scored against an agreed quality rubric (relevance, credibility, impact, newsletter fit), applied consistently, based on the actual full article rather than just a headline.
  3. 3Anything low-quality, paywalled, or without full readable content is automatically filtered out.
  4. 4The system ranks what's left into a shortlist of roughly the top 20 candidates.
  5. 5A human editor reviews the shortlist and makes the final call on the 10 stories that go into that week's issue. The system never picks the final list itself.
  6. 6The finished issue is assembled into a branded document with themed sections, sourced articles, and credibility notes for each pick.

Deliverables

  • Weekly published issues of The Shortlist (e.g. Issue 06, July 2026)
  • Article scoring and curation within GenServe.AI's rubric
  • Healthcare-audience framing for each featured story

Outcome

The Shortlist is a real, shipping publication, not a pitch. Each issue includes dated, sourced healthcare AI stories, framed for healthcare executives and researchers, with credibility notes that distinguish preprints from peer-reviewed research and clearly flag anything that couldn't be verified rather than guessing.

What I learned

  • Good AI-assisted curation still needs a clear, explicit rubric: consistency comes from the criteria, not from the AI alone.
  • A human editorial decision matters even when the filtering upstream is strong; the system's job is to narrow the field, not make the call.
  • Framing the same research differently for different audiences (researchers vs. healthcare executives vs. entrepreneurs) is its own skill.

Skills demonstrated

Healthcare AI research and evaluationEditorial judgmentAI-assisted workflow operationHealthcare-audience communicationTrend monitoring

Want to talk through how this applies to your team?