BioStack Raises $9.5M to Build a New Data Layer for Medical AI

San Francisco-based BioStack has raised a $9.5 Mn seed round led by Peak XV Partners, DCVC and Afore Capital, with Y Combinator, Emerson Collective and SV Angel also writing checks. The startup builds clinical data infrastructure that converts scattered, messy hospital records into clean datasets medical AI models can actually learn from.

What stands out is the scale BioStack is already operating at before this round even closed: the company has access to 100 million+ de-identified patient records spanning four countries the US, India, Brazil and Argentina. That’s a genuinely rare footprint for a seed-stage startup, and it’s likely what pulled in investors like Peak XV before the company had much else to show.

This raise fits a pattern investors are increasingly betting on funding the infrastructure behind medical AI, not just the AI models themselves.

The Details

• Round size $9.5 Mn, Seed
• Lead investors Peak XV Partners, DCVC, Afore Capital
• Also participating Emerson Collective, SV Angel, Verdict Capital, Streamlined Ventures, Y Combinator
• Valuation Undisclosed
• Founders Sanat Mishra (CEO) and Parth Patwa
• Footprint Hospital data partnerships across the US, India, Brazil and Argentina

100 Million Records, One Messy Problem

Patient data today sits scattered clinical notes in one system, lab results in another, imaging somewhere else entirely. No single hospital record tells the full story of a patient’s care over time.

BioStack’s core product stitches these fragments into longitudinal patient records linking notes, labs, imaging, medications, treatment history and outcomes into one continuous timeline.

Once de-identified, this becomes training material for medical AI models, drug development research, and clinical evaluation the kind of structured, high-quality data that’s historically been the biggest bottleneck in healthcare AI, not compute power.

Founder Credentials Signal Domain Depth

CEO Sanat Mishra isn’t a generalist founder pivoting into healthcare he conducted cancer genomics research at Stanford, Yale, Carnegie Mellon and the Max Planck Institute before founding BioStack.

He’s also worked directly with AI labs as an early researcher on reinforcement learning tasks applied to healthcare and biotech exactly the kind of background that explains why BioStack isn’t just selling static datasets, but also building simulated training environments for medical AI agents.

In India specifically, BioStack is already connecting cancer patients to an investigational vaccine program turning clinical data into a two-way value exchange between hospitals, patients and researchers.

Betting on Data, Not Just the Model

BioStack’s raise sits inside a broader investor shift: funding rounds like Amigo AI’s $11 Mn for training clinical agents and Snorkel AI’s $350 Mn for AI data infrastructure show capital moving toward the unglamorous plumbing layer of medical AI, not just flashy model announcements.

That’s a sound bet every medical AI lab needs clean data, but almost none of them want to build the compliance-heavy pipeline to get it themselves. BioStack is positioning itself as that layer for everyone.

What the Numbers Don’t Tell You

An undisclosed valuation on a $9.5 Mn seed with this investor list is normal, but the real risk here isn’t funding it’s trust. Healthcare data businesses live or die on provider relationships and regulatory compliance, not just dataset volume.

Having access to 100 million records means little if hospitals restrict usage rights or if privacy governance slows commercial deployment. The harder, slower-moving proof point will be whether BioStack’s datasets actually show up in published AI model benchmarks not just in its own pitch deck.

The Bottom Line

BioStack has the data scale and founder pedigree to matter the next test is whether hospitals keep the pipeline open as commercial pressure increases.