For Episode 34 (March 7, 2026), Sam built a digital twin dashboard for clinical trial site selection with our guest's company, Ryght AI, in mind. It simulates a network of trial sites and contrasts the usual manual strategy, heavy on tier-one academic centers, with an AI-optimized pick. The episode didn't cover the tools Sam used or how long it took, so this page covers what the dashboard does and how the expert reacted.
Ryght AI builds digital twins of clinical trial sites and uses AI agents to match trials to the sites and investigators most likely to enroll patients. Sam wanted to show the gap between picking sites by hand and picking them with AI, and opened at 03:12 by admitting it "may be totally off base."
Sam didn't name his tools on the show, so we won't. On screen, the dashboard had:
This dashboard simulates the clinical trial network using historical performance data and predicts enrollment velocity, diversity, and risk in real time.
— Sam Nadler, Episode 34
The delay and rankings are outputs of a demo, not real trial data.
Ryght AI's Chief Medical Officer, Dr. Chadi Nabhan, said it was "somewhat" how they think, then warmed up:
So I I like what you've built here. I think it's pretty easy and you certainly picked up on what we're doing.
— Dr. Chadi Nabhan, Episode 34
He explained why it matters at 06:12: trials fail when the drug doesn't work, when it's too toxic, or when the trial never enrolls. Ryght targets the third. He cited that 80% of trials run behind schedule and 50% of sites enroll zero to one patients. Ryght scores sites and sorts them into three tiers by how likely they are to enroll.
Rank sites per trial, not on a general prestige list. Dr. Nabhan's point at 07:57 was that the site ranked 20th on a list might be the best one for a specific trial, and the only way to find it is to match the trial's needs to each site's capabilities. Real sponsors add constraints, like needing European sites for a regulatory goal, and a fuller version would model that.
The census overlay turned out to be the most relevant piece: Ryght links its US digital twins to census data because trial enrollment often doesn't reflect the country's demographics. More on AI in healthcare, and for a related build, see the AI start-of-care packet generator. Jordan co-hosted.
In Sam's dashboard, it's a simulation of a trial site network built from historical performance data that predicts enrollment velocity, diversity and risk. Ryght AI builds digital twins of trial sites worldwide to match each trial with the sites most likely to enroll.
By matching a specific trial's needs against each site's capabilities instead of defaulting to famous academic centers. Dr. Chadi Nabhan said the site ranked 20th on a generic list could be the best one for a particular trial.
Dr. Nabhan gave three reasons: the drug doesn't work, it's too toxic, or the trial never enrolls enough patients. He said 80% of trials run behind schedule and 50% of sites enroll zero to one patients.
It compares site demographics against the US census so sponsors can pick sites that help trials reflect the real patient population. Dr. Nabhan said trial enrollment often skews heavily white and higher-income compared with the US.
