For decades, the US Medicare payment system paid for time. A doctor saw you, billed for the visit, and got paid. Everything that happened between appointments — the phone call that caught a warning sign early, the check-in that kept someone taking their medication, the coordination that got a vulnerable person connected to food assistance — had no billing code and no payment mechanism. It simply did not exist in the financial architecture of American healthcare.
That architecture just changed.
Starting July 5, a 10-year federal program called ACCESS begins paying healthcare organizations based on whether patients actually get healthier. Full payment requires hitting measurable targets: lower blood pressure, reduced chronic pain, improved depression scores. The program covers conditions that affect more than two-thirds of all Medicare patients, including diabetes, hypertension, kidney disease, and anxiety.
The significance of this is not clinical. It is financial. For the first time, an AI agent that calls a patient at 11pm, monitors their health data between visits, coordinates a housing referral, or handles intake and follow-up — all of that can now be the mechanism through which a healthcare organization earns its payment. The activity itself does not need to involve a human clinician.
The program was designed by former startup operators who now sit inside the government. The two architects, both appointed under the current administration, previously worked as a venture capitalist and a healthcare founder respectively. Their fingerprints are visible in the structure: outcome-based payments, patients enrolling directly without needing a doctor's referral, and deliberate competition between 150 participating organizations whose risk-adjusted results will be published publicly.
The payment rates are intentionally lean. That is the point. Organizations that still rely on large human teams to handle every patient interaction will find the economics do not work. The model is structured so that only AI-first operations — ones that have automated most of their patient touchpoints — can actually be profitable at the reimbursement rates on offer. The government has essentially written a payment rule that selects for automation.
The CMS Innovation Center's track record on cost savings is, to be honest, poor. Over its first decade, the center ran 49 evaluated programs. Only six generated significant savings, and only four were ever expanded nationally. The center spent roughly $7.9 billion and saved $2.6 billion during that period — a net increase to federal spending of around $5.4 billion. ACCESS is the same agency trying something very different under very different leadership, but the institutional history is real and worth holding onto.
The data risks are also real. Patients in this program are sharing some of the most sensitive information that exists — mental health details, housing instability, chronic disease management — with organizations feeding it into federal infrastructure that has a documented history of breaches. For the populations ACCESS specifically targets, many of whom are already vulnerable, that is not a theoretical concern.
What makes ACCESS different from past attempts is the investor capital building up behind it. Digital health funding reached $4 billion in the first quarter of 2026 alone, the strongest start to a year since the pandemic peak. The companies attracting the biggest rounds are the ones with contracted revenue, clinical evidence, and deep workflow integration — exactly the profile that ACCESS rewards. The program is not creating that investment wave, but it is the regulatory infrastructure the wave has been waiting for.
The companies most likely to win here are not the best-known names in health technology. They are the ones that spent years building trust with difficult patient populations, accumulated real clinical evidence, and now have the automated care infrastructure to operate at scale under thin margins. The wearable makers and consumer health brands that joined the first cohort are testing a very different hypothesis. Whether a fitness tracker meaningfully improves outcomes for a Medicare patient managing four chronic conditions at once is a question ACCESS will answer over ten years — with federal payment data attached.