Investing · How I think
Thesis
A lens, not a list of tips. These are the patterns I keep returning to across the memos — how I decide what's worth backing in health-tech and fintech, mostly across India and the US. Verdicts are moment-in-time; the ledger keeps the conviction honest.
What I look for
Distribution decides it
In health-tech and fintech the winner is usually whoever reaches users cheapest — owned networks, enterprise channels, cross-sell. I weight distribution and CAC over feature lists.
Underwrite the unit economics, not the GMV
I reward durable margins over scale bought with cash. PharmEasy's roll-up is the cautionary case; the profitable players are the counter-example.
In health AI, clearances are the moat
Regulatory approval and clinical validation are harder to copy than models — it's why a cleared diagnostics company screens differently than a wrapper.
Embedded beats standalone
Products that live inside the clinician or user workflow — and compound data as they run — are stickier than destinations you have to remember to visit.
Reach the underserved without burning the model
I'm drawn to businesses that extend care or capital to rural and low-resource users while keeping unit economics intact — the problem I worked on directly at Juno.
Price is part of the thesis
A company I admire can still be a PASS at the wrong valuation. Quality and entry price are separate decisions.
The verdict ladder
See the calls →A clear distribution or regulatory moat, a credible path to durable economics, and workflow lock-in.
Abridge · Qure.ai · MediBuddy
A strong asset where price or near-term execution caps the upside.
Tata 1mg · Groww · Innovaccer
A thesis I find interesting but can't underwrite yet — unproven monetization or a catalyst not in hand.
Healthify · CRED
A company I respect but can't underwrite at the current price or structure.
OpenEvidence · PharmEasy
Where I focus
India and the US.
- Telemedicine
- Diagnostics
- E-pharmacy
- AI healthcare
- Health insurance
- Rural healthcare
- Mental health
- Women's health
- Hospital infrastructure
- Consumer fintech
How I keep it honest
Date the call
Every prediction gets a probability and a resolution date, then reality grades it. No after-the-fact narrative.
Score, don't assert
Memos rate five dimensions on a five-point scale, and the calibration ledger (0.134 Brier) keeps conviction from drifting into bravado.
Verdicts are moment-in-time
Buy / Hold / Watch / Pass describe the call at today's price and moment — not a permanent judgment of the company.