Angle: Rebuild the Guts

Status: Noodling Date: Feb 13, 2026

The Thesis

Healthcare infrastructure is worth $160B in enterprise value and it's rotting from the inside. The companies that run the core systems — TriZetto for claims, ZeOmega for care management, Change Healthcare for clearinghouse, Epic for EHR — are legacy stacks built 10-20 years ago. They work. Barely. And nobody inside these healthcare orgs knows how to rebuild them.

a16z mapped the entire infrastructure taxonomy. The payer-side Care Management box — where utilization management lives — has one company (ZeOmega). ONE. Compare that to provider-side Revenue Cycle, which has 9+. That's the most underbuilt, underserved quadrant in a $160B market.

Meanwhile, the orgs themselves are paralyzed:

Summa Health can't provision a single LLM (Dan's Percepta/GC experience, Feb 2026)
Premera told us enterprise product procurement takes 6-9 months through security and review
Payer IT spend is 3.5% of revenue — half of what other industries spend

These orgs know they're behind. Their stock prices reflect it (Thomas, Feb 13 call). They're ready to let someone in to look at their tech stacks. But the legacy vendors can't innovate (they ARE the legacy), the big consultancies don't build AI-native, and the orgs can't do it themselves.

Someone has to go inside and rebuild the guts.

Why Now

Julie Yoo's leapfrog thesis (a16z): healthcare's low software adoption is now an asset. No sunk cost bias. Other industries spent hundreds of billions on enterprise software and now face rip-and-replace decisions. Healthcare goes straight from fax machines to AI.

"In healthcare, the question is simply — should I throw more bodies and fax machines at the problem, or should I jump to adopt AI without the baggage of last-generation workflow apps getting in my way?"

The payers are waking up to this. Premera is inviting us in for 3-6 months of "PhD work" to study their systems. That's not a consulting gig — that's a healthcare org saying "we know we need to rebuild and we don't know how."

Mello's stats back it up: 84% of large insurers using AI for some purpose, but governance is a mess — 40% have no committee reviewing AI tools, 25%+ don't test for bias or track accuracy. They've adopted AI chaotically because they had no one to architect it properly.

The a16z Map

From their infrastructure taxonomy:

Payer side — where we sit:

Care Management (UM, pop health): 1 company (ZeOmega). Desert.
CAPS (claims, eligibility, payments): TriZetto (Cognizant), HealthEdge. Legacy.
Risk Management: Cotiviti, Verisk. Adjacent, not our fight today.

Provider side — for comparison:

Revenue Cycle: 9+ companies (Change, Waystar, R1 RCM, Availity, etc.). Crowded.
Clinical Systems: Epic, Oracle Cerner, Nextgen, etc. Mature.

The emptiness of the payer Care Management box IS the opportunity. We're not competing in a crowded category. We're filling a vacuum.

The Play

Get inside. FDE consulting engagements at payers (Premera is first). Get permissioned like employees. Map their systems, understand their data flows, learn where the legacy breaks.
Rebuild from within. Replace ZeOmega-era workflows with AI-native architecture. Auto-auth, appeals processing, concurrent review — built on modern stacks, not 15-year-old platforms.
Extract reusable product. Every rebuild produces IP that transfers to the next payer. The consulting engagement funds the R&D. The product emerges from the work.
Expand across the payer stack. Care Management box first. Then touch CAPS (claims integration), Risk Management (payment integrity), Provider Network (contracting). Move across the taxonomy.

Why Not Just Sell Software

Because they can't buy it. Dan said it plainly (Feb 13): "The amount of information you get as a contractor who's hired versus as a vendor — it's 180. As a vendor, they're guarded and expect you to have all the answers."

You have to be inside to rebuild the guts. You can't do it from outside the walls. The FDE model isn't a compromise — it's the only way in.

And once you're in, you see everything. You know where the real problems are. You build what they actually need, not what you guessed from a demo. That's how you build infrastructure that sticks.

How It Connects

Vibe Coder Army = the workforce that goes inside and does the rebuilding
Clearing House = the end-state infrastructure that emerges when you've rebuilt both sides
Premera engagement = the first rebuild, the training ground, the proof of concept
Dan at Summa = seeing the same problem from the provider side (they can't even provision an LLM)
Team of 5 = the legacy system integration expertise (HL7, ADT, EHR connectors) needed to actually interface with the old guts while building the new ones
Mello's article = the regulatory pressure that forces these orgs to rebuild responsibly rather than just slapping AI on top of broken processes

The Line

"Healthcare infrastructure is worth $160B and it's crumbling. The payer side is the most underbuilt quadrant. These orgs can't rebuild themselves — Summa can't even provision an LLM. We go inside, we rebuild the guts, and we extract product along the way."

Open Questions

How do you avoid being "just a consulting firm" in the investor's mind? The product extraction piece is critical — without it, this is Accenture.
Is the a16z taxonomy the right framing for all investors, or just for a16z? Some VCs might not know Julie Yoo's work.
How long before the first rebuild produces transferable product? 6 months? 12?
Does the ZeOmega comparison hold up? Are they really the only player, or are there startups we're missing?
What's the expansion path beyond Care Management? Do investors want to see a multi-box vision or focus?