A quiet arms race is underway:
OpenEvidence currently sits at a $6 billion valuation, betting that clinicians want live evidence synthesis powered by large language models (LLMs) trained on structured clinical data, for free.
UpToDate is building its own “generative” overlay, while competitors like Elicit and Consensus scrape and summarize new papers in real time.
Behind the scenes, every major EHR vendor is exploring an AI layer that pulls research straight into the chart.
But here’s the catch: These systems are trained on conventional medicine. They excel in cardiology, oncology, and internal medicine. Not in the frontiers where functional, integrative, and longevity practitioners live.
When you search for insights on peptides, senolytics, methylation clocks, or stool-based biomarkers, the response is usually a shrug or a red flag.
Like Looking in a Mirror
AI mirrors the medical mainstream because that’s the data it’s built on, and the data most guidelines reinforce.
The reality? Clinical innovation still moves faster than clinical consensus. Case in point: it wasn’t until June 2025 that the first guideline on lifestyle interventions for treating and reversing type 2 diabetes and prediabetes was published by the American College of Lifestyle Medicine (ACLM).
Traditional medicine writes guidelines to standardize care. Longevity medicine builds frameworks to individualize it. That shift from population averages to human complexity pressures a system already stretched thin.
A daily surge of omics data, novel biomarkers, and experimental interventions now moves faster than any one clinician or journal can process. AI isn’t just noise management anymore. It’s how the most advanced practices are beginning to filter, prioritize, and translate the most clinically relevant studies into personalized precision medicine in real time.
Closing the Loop: What Works (and What Still Doesn’t)
Here’s what we’re seeing across top practices:
What’s working now:
Using tools like OpenEvidence, Elicit, or Consensus to triage emerging data, filtering by study design, N, and recency.
Embedding evidence snippets directly into your care plans and patient education.
What still doesn’t work:
ChatGPT on its own (too general, too willing to hallucinate PubMed citations - Deep Research mode is a different story).
Using LLMs to “interpret” papers without grounding or verified data sources.
Over-reliance on models that aren’t finessed for clinical nuance or bias control.
📙 Your Homework: Be the Human in the Loop
You don’t need enterprise-grade tools or technical expertise to close the gap between current AI and what you’re doing at your clinic. Try this practical workflow:
Create a standing research query: Ask ChatGPT or Claude to monitor new studies on a precise clinical question (example prompt: “monitor new studies on intermittent fasting + insulin sensitivity in perimenopausal women”). Save that as a reusable prompt.
Pair it with a structured evidence filter: Use Elicit, FunctionalMind, or Consensus to cross-check claims with real studies and extract sample size, effect magnitude, and study type.
Read the relevant studies: Ground yourself in the primary evidence so you can judge clinical relevance and implement only what fits your competence and practice setting.
Close the loop with AI summarization: Ask your model to summarize the top three changes to your clinical protocol this quarter based on new evidence.
These micro-loops compound. Within weeks, you’ll have a constantly updating evidence map instead of scattered PDFs.
ICYMI: Jona x Vibrant Webinar
To learn more about Vibrant, the AI-powered, all-in-one practice platform, schedule some time with our team today.
This Week in Clinical AI
The quiet war over who will become the next “Google of medicine” is underway. OpenEvidence announced another $200 million fundraising round just three months after its last $210 million round. Meanwhile, UpToDate announced its long-awaited AI update.
A smartphone-based digital ruler uses AI to measure strabismus. The AI tool, which works from phones, delivers a Digital Ruler of Strabismus (DRS) that accurately measures strabismus from patient-recorded videos. This offers a scalable, clinic-adjacent method for early detection and longitudinal monitoring in both pediatric and adult populations.
K-Dense, an AI-driven transcriptomic aging clock, estimates biological age with high accuracy. It also signals when predictions may be less reliable. It offers a glimpse into how future clinical tools might integrate uncertainty-aware biomarkers to personalize aging assessments and guide precision interventions.
👋 Welcome New Readers
The Modern Clinician is written for functional, integrative, and longevity-focused physicians who want to scale their impact and deliver cutting-edge care.
If you liked this one, share it with a colleague! We appreciate you spreading the word.
To learn more about the why behind this newsletter, start with our first post introducing The Modern Clinician.