Leapfrogging Legacy Infrastructure With AI-Native Healthcare Services

The healthcare industry is moving from a paper-heavy, legacy-constrained system to an AI-native infrastructure. As a16z General Partner Julie Yoo explains, this shift is systemic: by skipping the middleware layers other industries spent decades building, healthcare can move directly toward agentic AI. This transition is driven by consumer frustration, high costs, and the arrival of AI tools that offer immediate value to clinicians. For founders and investors, the opportunity lies in solving high-stakes, regulated problems at a fraction of historical costs. This creates a window where services that solve immediate pain points can build market moats while the industry's payment and delivery models change.

The leapfrog advantage of legacy debt

Most industries spent the last two decades building rigid software layers that now function as technical debt. Healthcare avoided this investment, which puts it in a position of strength. Because it lacks the sunk costs of enterprise software, it can adopt agentic AI directly.

Yoo notes that while other industries face the difficult process of replacing old systems, healthcare can bypass those steps. This is already happening. Unlike past mandates, such as government-subsidized electronic health records, today's AI tools like automated scribes are being adopted by clinicians because they provide immediate relief from administrative work.

"Healthcare for better for worse did not spend that amount of money... we only had really like the ERP layer with EHRs and then labor. Like we were just throwing bodies at every problem. And so in some ways, we have less of a sunk cost bias as an industry to say, okay this new technology paradigm has come along and we can just leap right to it."

-- Julie Yoo

The shift from payer-first to consumer-first

Conventional wisdom has long held that healthcare innovation must design for insurance companies or hospital systems, with the patient as an afterthought. That is changing. As high deductibles force consumers to pay for services out of pocket, individuals are choosing better, faster, and cheaper care.

This creates an advantage for startups that deliver consumer-grade experiences. By using AI to lower delivery costs, these companies can offer services at prices that disrupt the traditional, bloated system. This is about shifting the power balance in a third-party payer system where costs were once hidden.

"If you had come to any investor, let's call it seven years ago and pitched a consumer health business, you would have been kicked out of the room because there was no viable business model... Now that all that's changed and some of the fastest growing companies that we see in HealthTech are cash pay you just pay out-of-pocket for a service at a disruptively low price."

-- Julie Yoo

Solving the last mile of AI-native care

Generalist AI tools provide instant access to intelligence, but intelligence alone is not healthcare. The opportunity for founders lies in the last mile: verifying diagnoses, prescribing treatments, and providing longitudinal care.

The most successful companies will be AI-native and AI-proof. This means building full-stack, regulated entities that use AI to reach cost structures incumbents cannot match. By combining AI efficiency with the requirements of a licensed practice, these companies build defensible moats. This requires patience, as these businesses must navigate the complexities of healthcare infrastructure, but the result is a direct, long-term relationship with the patient as their AI doctor for life.

Key action items

• Prioritize AI-native over AI-enabled: Focus on building full-stack services where AI is the core delivery mechanism, rather than a wrapper around legacy workflows.
• Target the cash-pay consumer: Identify services currently trapped behind insurance friction that can be delivered at a low price point. This is an immediate opportunity over the next 12 to 18 months.
• Solve the last mile of intelligence: Do not just build diagnostic tools; build the infrastructure to act on that data, including prescribing, referring, and longitudinal monitoring.
• Exploit the lack of sunk costs: Avoid building middleware that mimics legacy software. Design workflows that assume an agentic AI future.
• Build for longitudinal data: Invest in collecting the narrative arc of patient health, which is currently missing from fragmented electronic health records. This is a long-term investment of 3 to 5 years that will power the next generation of medical AI.
• Leverage high-acuity robotics: Explore the intersection of AI and hardware in settings where physical intervention is required but labor is scarce.