In 2026, we have entered the era of trillion-dollar biology. Companies like Xaira can design custom proteins in seconds. Breakthroughs in CRISPR and GLP-1 therapies are reshaping what is possible in human health.

Yet the software infrastructure that supports this miracle science and care delivery remains stuck in the late 1990s.

The irony is hard to ignore. While we can edit genomes, a ward nurse is still forced to note vitals on a scrap of paper before manually entering them into a legacy EHR (electronic health record).

At Adeline, we have spent months analysing the manual absurdity of the current patient journey. Our conclusion? We are backing founders building the future infrastructure of healthcare at pre-seed and seed stages (if this is you - pitch us.)

Why? The data is clear: Health AI is no longer a “vitamin” or nice-to-have. It is the only realistic path to recovering the estimated $140 billion in care capacity lost each year to documentation and administrative burden.

The diagnostic gauntlet: a story of "hops"‍

The modern patient journey is a series of disconnected hops. Whether in London or New York, the process is rarely digital-first. It typically begins with a phone call to secure a GP or primary care appointment, followed by a scheduling marathon.

If a diagnostic test is required, the situation becomes more complex. Post-scan, the data often enters what we call a three-party data swamp between the GP, the diagnostic centre, and the specialist. Because these systems are siloed, information is shared through manual uploads, scanned documents, and fragmented permissions.

This is not just an administrative inconvenience. It is a clinical risk. Studies indicate that clinicians with access to integrated data make 37% fewer diagnostic errors. 

The clinician tax: a bottom 9% experience‍

The friction extends to the point of care.

Nurses still record vitals by hand. Verbal handovers remain common because multi-billion-dollar EHR systems function primarily as Systems of Record, not Systems of Action.

Physicians rate EHR usability in the bottom 9% of all software systems globally. Doctors spend roughly 37% of their workday interacting with the EHR. That is more than a third of their time spent in software that many find frustrating and unintuitive. 

Studies consistently show that current EHR design contributes to inefficient workflows and clinician burnout.

This is not a marginal issue. It is a structural tax on clinical capacity.

The insurance bottleneck: manual by design‍

For patients with private insurance, the hops multiply.

The patient becomes the project manager of their own care, calling providers to verify coverage, confirming claims eligibility, and manually relaying updates back to the insurer. Documentation is uploaded, downloaded, and re-uploaded across disconnected portals.

Meanwhile, the insurance provider often has limited real-time visibility into what is actually happening clinically.

Building the nervous system of healthcare

We are excited by founders building the future infrastructure of healthcare

The ‘trust gap’ in health tech is narrowing because the value of AI-native automation is now clear and undeniable from the perspective of both customers and end users.

We are looking for Health AI startups that go beyond surface-level tools and wrappers to fundamentally rebuild the healthcare tech stack. 

We are specifically hunting for founders solving for 2 main points of failure: 

1. The data infrastructure gap
2. The care orchestration gap 

We are seeing a surge in startups like Abridge, Nabla and Tandem Health that use ambient listening to capture the natural conversation between a doctor and a patient. The goal is simple: return the doctor to being a healer, not a data entry clerk.

The most impactful solutions operate across multiple users and devices, interoperating seamlessly on desktops, tablets, and mobile platforms, while bridging both legacy and modern systems. Reducing adoption friction is challenging, especially with high capex costs and multi-year contracts, but it is exactly what makes these solutions transformative.

We would like to see new technology and solutions that take this a step further, such as: 

Workflow orchestrators: Agentic systems that handle the hops autonomously, including referrals, insurance checks, and follow-ups. Imagine an AI that hears, “I’m going to refer you to cardiology for a stress test,” and automatically drafts the referral, initiates the insurance pre-authorisation, and surfaces a one-click workflow for all parties involved.

The "system of action" overlay: Software that renders Windows 2000-era EHRs effectively invisible to clinicians by providing a modern, intelligent interface that actively supports decision-making. Instead of a nurse writing vitals on paper, AI-enabled medical instruments can stream data directly into the EHR through an orchestration layer that validates and structures the information in real time.

If you are building in this space, we would like to talk: we are actively investing in pre-seed to seed startups solving clinical workflow and data silo challenges.