The Body Data Pipeline
Dan Peguine@danpeguine
−3.9%
since Sep 30
The idea
I believe AI can already diagnose better than most doctors, so the bottleneck isn't intelligence, it's getting readable data out of the human body. This basket leans hardest into continuous sensors people wear daily, backs that with blood and lab testing, and adds smaller bets in genomics, imaging, and agent-facing consumer rails. It's built concentrated on purpose since I think adoption inflects within the next year or two, with 10% cash kept aside to add if things get volatile. I'd rethink this if consumer adoption of these tools stalls out instead of accelerating.
What it owns
- Real-time data flowing from the body is the foundation for any agent to monitor health continuously, and these companies already ship sensors at scale to millions of people. · DexcomIt streams blood sugar data continuously in a format an AI agent can read all day long, making it the core sensor for monitoring the body's most accessible signal in real time., AppleThe Apple Watch sits on hundreds of millions of wrists as a universal health sensor, and opening its data stream to agents would create the broadest harness for continuous body monitoring ever built., Abbott LaboratoriesIts Libre sensor already ships at consumer scale, building the mass-market data pipe that an agent needs to read glucose outside the clinic and without a doctor's order., iRhythm TechnologiesThe wearable patch records heart rhythm continuously and uses AI to detect patterns that spot checks miss, positioning it to capture share as cardiac care shifts from episodic to continuous.32%
- Blood panels reveal disease signals that wearables cannot, and agents ordering frequent screening means far higher testing volumes through the national lab networks. · Quest DiagnosticsIt runs one of the two national lab networks through which nearly every blood draw flows, and its direct-to-consumer panels let agents order deeper data without involving a doctor., Guardant HealthIt reads cancer signals from blood alone, making it the asymmetric bet on whether high-stakes screening like blood-based cancer detection becomes routine rather than exceptional., LabcorpIt is the other national lab rail, so if agents start ordering blood work at scale, test volume flows through here as the infrastructure layer supporting routine screening., NateraIt reads genetics from blood across prenatal, cancer, and transplant signals, making blood-based genetic data a routine input rather than a specialty test as volumes compound.25%
- Genetic and molecular data are the deepest signals an agent can order, and the falling cost of reading DNA makes routine sequencing and cell-level analysis increasingly practical. · Tempus AIIt pairs AI models directly with a deep clinical dataset, making it the clearest way for an agent to order diagnosis-grade molecular and genomic data from the same platform., IlluminaIt sets the cost curve for reading genomes, and genetic data only becomes a standard input if sequencing falls cheap enough that every person's agent has a genome on file., 10x GenomicsIt reads the body at single-cell resolution, making it an asymmetric bet on whether deep omics moves from research into routine clinical use during the next few years.16%
- An agent's health recommendation only matters if it can reach a person as a prescription or test kit, and these distribution channels win by processing volume at low cost. · Hims & Hers HealthIt runs the consumer front door where an agent's recommendation converts into a prescription and delivery, capturing the volume play in turning digital advice into actual medicine.9%
- Scans find what blood tests and sensors miss, and preventive screening driven by agents means more imaging per person over a lifetime. · GE HealthCareIt builds the MRI and CT machines that capture whole-body imaging, and consumer demand for preventive scans would pull equipment sales through to the hardware makers., PhilipsIt spans both imaging scanners and bedside monitors, so it profits whether body data comes from a scanner or a sensor, as long as monitoring keeps shifting toward continuous and connected.9%
- Cash10%