The Last Mile Problem in Health Care with Angela Adams

Every day across American hospitals, clinicians detect problems they cannot always ensure get acted upon. A radiologist flags an abnormality, a lab result returns days after a patient has been discharged, a pathology report sits in a system no one is actively monitoring. The diagnosis exists. The follow-through does not.

Angela Adams, founder and CEO of Inflow Health, calls this the last-mile problem in health care. It is the subject she has spent years working to solve, and on a recent episode of the Scopewell podcast, she laid out why the gap is structural, how her company evolved to address it across multiple care settings, and what it actually takes to change a system that, in her words, was designed to fail.

A personal loss, a systemic problem

Adams trained as an ICU nurse and came to entrepreneurship through a painful detour. A close friend died after a missed radiology follow-up, a preventable outcome that Adams describes as the direct motivation for founding Inflow Health. "I don't want this to happen to anybody," she said. "It's a 100% preventable thing."

She chose radiology as the company's starting point because imaging sits at the front of nearly every clinical pathway. Patients arrive at emergency departments, primary care offices, and specialist clinics, and in each case imaging is typically ordered to establish what is wrong. Yet radiology reports are dense, physician-to-physician documents, and the data inside them is rarely tracked or acted upon in any systematic way.

That gap between the report and the response is where patients fall through.

Systems built for billing, not for care

Adams traces the root of the problem to the way electronic health records were designed. She points to the meaningful use era, when the health system digitized the medical record and built out robust billing infrastructure, but did not build tools to help clinicians access and act on data in the moment care decisions are made.

"Health care doesn't struggle with knowing what to do," she said. "It struggles with actually getting it done." The result is a system where documentation is thorough and billing is automated, but the quality of care in between those two functions is largely ungoverned.

From her years as a bedside nurse and through root-cause analyses of adverse events, Adams observed repeatedly that negative patient outcomes were not failures of individual clinicians. They were failures of system design. Her goal with Inflow Health is to build the infrastructure that fills the space between detection and action.

From radiology to a platform

Inflow Health began by automating follow-up workflows for abnormal radiology findings. But as the company deepened its relationships with health system clients, the same pattern appeared in other departments. Cardiology teams managing heart failure patients, laboratory teams processing microbiology results that return days after a patient's discharge, and pathology departments dealing with complex report structures all faced identical coordination failures.

Adams credits close client engagement for surfacing the broader opportunity. Rather than defending her original product scope, she says she monitors continuously whether the value her company delivers matches what clients actually need. Clients began telling Inflow that the radiology problem was not unique to radiology.

"The system doesn't fail at detection," Adams said. "It fails at what happens next." That insight reframed Inflow Health's mission from a radiology-specific tool into what Adams now calls a last-mile care orchestration platform.

A four-pillar filter for the product roadmap

As AI tools lower the cost of building features, Adams argued that the harder discipline is deciding what not to build. She uses four value pillars to vet every product decision. The idea must address a meaningful patient safety gap. It must mitigate liability for the health system and support better clinical decision-making. It must drive its own return on investment, without depending on future reimbursement changes or regulatory approvals. And it must be staff-neutral, meaning it cannot require health systems to hire additional people to realize its value.

She illustrated the framework with a concrete example. Multiple clients asked Inflow to translate radiology reports from medical jargon into plain patient language, pointing to competitors already offering that feature. Adams declined to copy the approach. Instead, she questioned whether a text translation was actually how patients consume information in 2026, concluding that short, interactive video explanations paired with a guardrailed large language model would serve patients better.

The result is a patient-facing layer where a video explains findings in accessible terms, an LLM answers follow-up questions using the patient's own report data, and a scheduling prompt turns understanding into action. "I didn't just copy what else was out there on the market," Adams said. "I rethought how we as consumers are doing things and applied that towards the problem."

Why health system sales remain a long game

Adams is candid about the friction involved in selling to hospitals and health systems. The challenge, she argues, is not that buyers fail to see value. It is that no single stakeholder owns a clinical problem end to end. Without clear ownership, prioritization is difficult, and approval processes can stretch across many internal committees where a single objection is enough to halt a deal.

Staffing shortages compound the problem. Even when a health system wants to adopt a new tool, the clinical and operational bandwidth to implement it is often not there. Adams frames care orchestration as a response to this reality: the platform is designed to absorb coordination work rather than add to the burden of already stretched teams.

She describes care transitions as the highest-risk moments in a patient's journey. When a patient moves from the emergency department to surgery, or from inpatient care back to a primary care physician who may not even know a hospitalization occurred, information is lost and accountability becomes diffuse. "Who owns James?" she asked, using the host as an example. That question, she said, is what care orchestration is built to answer.