The Bald Informaticists
An ED patient's chart is often longer than a novel — and it keeps rewriting itself while a physician is trying to read it before walking into the room.
Core trick
Rules engine flags new results instantly with chart-finding and guideline citations, while LLM agents write delta briefs as the chart changes.
Description
An ED patient's chart is often longer than a novel — and it keeps rewriting itself while a physician is trying to read it before walking into the room. Physician Co-pilot puts an AI layer directly in the chart to fix that: an instant, rules-based clinical decision support engine flags what matters the moment new results land, with every recommendation grounded twice — the exact chart finding and the guideline behind it — plus one-click, pre-filled actions to act on it immediately. Underneath, a multi-agent LLM pipeline synthesizes a plain-language pre-visit brief and delta updates as the case evolves, personalized to each physician's preferences over time. In our demo, an ordinary chest-pain workup evolves live into a confirmed aortic dissection: the system catches it, orders the confirmatory scan, and hands the physician five ready-to-send actions the instant it's confirmed. Although this demo shows only one specialty, this can be architected to scale to all fifty specialties Abridge already supports across settings. Now that's scale.
Stephonomon/abridge-previsit-copilot★ 10 forksTypeScriptCreated Jul 18, 2026 · last push Jul 19, 2026
README excerpt
# Physician Co-pilot *An agentic clinical decision-support demo for the Emergency Department, built at the [Abridge x Anthropic "Future of Agentic AI in Healthcare" hackathon](https://cerebralvalley.ai/e/abridge-hackathon), San Francisco.* **🎬 Watch the demo:** [youtu.be/TSwHBSF12vo](https://youtu.be/TSwHBSF12vo) --- ## Why we built this Abridge's CEO, Shiv Rao, opened the hackathon by naming the problem every clinician already knows in their bones: **critical patient information is fragmented across the chart, hard to find under time pressure, and easy to miss** — a stray line in a faxed outside note, a med buried three tabs deep, a trend that only means something once you connect two results that landed an hour apart. The challenge to 150 hackathon participants: build something with real-world impact a healthcare team could actually use on Monday morning. [Stephon Proctor](https://github.com/Stephonomon) and John Lee, MD (Emergency Medicine) spent the day building an answer to that problem for the ED, in 12 hours: a copilot that lives inside the chart, reads everything the physician doesn't have time to, and surfaces what actually matters — the moment it matters — with the reasoning shown and the next action one click away. ## What it does Physician Co-pilot pairs two complementary AI systems inside a mock Epic-style EHR: - **Abridge AI — instant clinical decision support.** A deterministic rules engine that matches live chart findings against an authored knowledge